Opinion
KENNARD, J.A jury found that plaintiff employee, Charlene J. Roby, was wrongfully discharged based on her medical condition and related disability. The jury found both harassment and discrimination, and it awarded $3,511,000 in compensatory damages and $15 million in punitive damages *693against the employer, as well as $500,000 in compensatory damages and $3,000 in punitive damages against the supervisor who was responsible for the harassment. Defendants appealed.
The Court of Appeal concluded that some of the noneconomic damages awards overlapped one another, and that the evidence was insufficient to establish harassment. It ordered the trial court to enter judgment in favor of the supervisor, and it ordered the trial court to modify the judgment against the employer to reflect a reduction of compensatory damages to $1,405,000. The court further concluded that the award of punitive damages against the employer exceeded the federal constitutional limit, and it ordered a reduction of punitive damages to $2 million. The Court of Appeal then affirmed the judgment as modified.
We granted plaintiff’s petition for review, which raised three issues. First, did the Court of Appeal err in concluding that some of plaintiff’s noneconomic damages awards overlapped one another? Second, did the Court of Appeal err in allocating plaintiff’s evidence between her harassment claim and her discrimination claim, and, based on that allocation, in finding insufficient evidence to support the harassment verdict? Third, did the Court of Appeal err in concluding that the punitive damages against the employer exceeded the federal constitutional limit?
With respect to the first issue, we conclude that the jury’s noneconomic damages awards are hopelessly ambiguous. In a letter to this court and again at oral argument, plaintiff’s counsel stated that plaintiff preferred to concede this issue rather than face a new trial, and defendants accepted this concession. Therefore, the validity of the Court of Appeal’s conclusion that some of the noneconomic damages awards overlapped one another is no longer in dispute. With respect to the second issue, we conclude that the Court of Appeal erred in allocating the evidence between the harassment claim and the discrimination claim, and we reject its determination that the record included insufficient evidence to support the harassment verdict. With respect to the third issue, we agree with the Court of Appeal that the punitive damages exceeded the federal constitutional limit, but we disagree with the Court of Appeal on the amount of this limit. We hold that in the circumstances of this case the amount of compensatory damages sets the ceiling for the punitive damages.
I
A
This matter is before us on appeal from a judgment in favor of plaintiff Charlene J. Roby, after a jury trial. In summarizing the facts, we view the *694evidence in favor of the judgment. (Bickel v. City of Piedmont (1997) 16 Cal.4th 1040, 1053 [68 Cal.Rptr.2d 758, 946 P.2d 427].)
Roby worked for defendant McKesson Corporation from 1975 until 2000. McKesson is a distributor of pharmaceutical and health care products, supplying both hospitals and pharmacies. At the end of her career with McKesson, Roby was a customer service liaison at a local distribution center, processing forms and handling customer problems related to product delivery. She did her job well and received favorable performance reviews. Starting in 1997, Roby began experiencing “panic attacks” that temporarily (and on short notice) restricted her ability to perform her job. During a panic attack, Roby suffered heart palpitations, shortness of breath, dizziness, trembling, and excessive sweating.
In 1998, McKesson instituted a complex attendance policy. The policy put particular emphasis on 24-hour advance notice for all absences, including medical absences. An absence without notice that lasted more than half the scheduled work shift was denominated an “occasion,” and two incidents of tardiness or early departure also counted as an “occasion,” but the term “occasion” referred to a period of absence that began without the required 24-hour notice, not to each day of absence. For example, if an employee suddenly became ill and was absent (without 24-hour advance notice) for three consecutive workdays, the three-day absence would be deemed a single occasion.
McKesson imposed progressive levels of discipline based on the number of occasions an employee accrued in any 90-day period. The discipline proceeded in a “3-3-2-2 sequence.” Three occasions in any 90-day period would result in an oral warning, and an additional three occasions in any subsequent 90-day period would result in a written warning. After the written warning, two more occasions within any 90-day period would result in a final written warning. After the final written warning, two more occasions within any 90-day period would result in termination of employment.
An employee would repeat a level in the sequence (rather than progressing to the next, more severe disciplinary level) if the. employee became eligible for discipline but had received no discipline during the preceding six months. If the employee became eligible for discipline but had received no discipline during the preceding 12 months, the level of discipline would be one level below the level last imposed (though the minimum discipline was always an oral warning). For example, if an employee received a final written warning but then received no discipline for six months before again becoming eligible for discipline, a second final written warning would be issued. If the same employee had received no discipline for 12 months before again becoming *695eligible for discipline, there would be a written warning (nonfinal), rather than a final written warning.
McKesson’s attendance policy operated to the disadvantage of employees who, like Roby, had disabilities or medical conditions that might require several unexpected absences in close succession. Roby accrued a large number of occasions in a relatively short time span, and most of them were directly or indirectly related to her panic attacks. Roby’s supervisors— including defendant Karen Schoener — were aware that Roby suffered from these unpredictable panic attacks and that many, if not all, of her absences without notice resulted from this condition.
Roby struggled to overcome her disability and to improve her attendance record, but after Schoener took over as Roby’s immediate supervisor, Roby’s frequent absences resulted in tension between them. Compounding this problem, Roby’s medication caused her body to produce an unpleasant odor, and in connection with her panic attacks she also developed a nervous disorder that caused her to dig her fingernails into the skin of her arms, producing open sores.
Schoener made negative comments in front of other workers about Roby’s body odor, although Schoener knew from Roby that medication was causing the odor. Schoener also called Roby “disgusting” because of the sores on her arms and her excessive sweating. Schoener openly ostracized Roby in the office, refusing to respond to Roby’s greetings and turning away when Roby tried to ask questions, and Schoener made a facial expression of disapproval when Roby took rest breaks because of her panic attacks.
Schoener also ignored Roby at staff meetings, and she overlooked Roby when handing out specialty food items, holiday gifts, and travel trinkets, although Schoener regularly gave these small gifts to the other employees on her staff. Schoener effectively excluded Roby from office parties by designating her to cover the office telephones. In addition, Schoener frequently reprimanded Roby in front of her coworkers. She spoke about Roby in a demeaning manner and openly belittled Roby’s contribution to the company, calling her job a “no brainer.” According to the testimony of one coworker, when Roby would telephone the office in the morning to report that she would be absent, Schoener “would always make this announcement that was degrading; say, ‘Charlene’s absent again’ — you know — that type of response.” Roby’s complaints to more senior managers about Schoener’s conduct went unanswered.
In early 1999, Roby accrued three occasions within a 90-day period. Although Roby told her then supervisor (not Schoener) that these absences *696were due to her medical condition, she nevertheless received an oral warning on April 2, 1999. By June 8, 1999, Roby had accumulated three and a half more occasions. She again informed her supervisors (who at this time included Schoener) that her absences were because of her medical condition, but she nevertheless received a written warning.
Roby then had two more occasions — July 27 to 28, 1999, and October 18, 1999. She gave her supervisors (including Schoener) a note, signed by her psychiatrist, stating that her July 27 to 28 absence was necessitated by a medical condition that was not contagious. Nevertheless, on October 22, 1999, Roby received a final written warning. She responded by telling Schoener that all her absences were because of her panic disorder.
After the final written warning, Schoener told Roby that if she had no further occasions before January 2000, she would have a “new start.” Roby interpreted that statement to mean that she would clear her poor attendance record if she succeeded in having no occasions between then and January. Roby met that goal and went to Schoener to express her delight, but Schoener said nothing.
In 2000, Roby had two more occasions, one on February 25, because of laryngitis, and the second on April 11, because of a panic attack. On April 13, 2000, two of McKesson’s local managers (including the head of the distribution center where Roby worked) met with Roby and told her that she had abused the company’s attendance policy and was subject to termination. Roby protested, explaining that in 1999 Schoener had assured her a “new start” if she lasted until January 2000 without any occasions. Roby also asserted that McKesson had applied the attendance policy unevenly, overlooking instances when other employees were absent without notice. Roby requested that her occasions be retroactively reclassified as protected medical leave under the Family and Medical Leave Act of 1993 (29 U.S.C. § 2601 et seq.) (FMLA), but the documentation she relied upon in support of this reclassification was limited to the brief medical notes that were already in her personnel file. These notes stated only that Roby “has been diagnosed with panic disorder,” that it is “not contagious,” and that the “[pjanic episodes have been stabilized [with medication].” The notes did not describe the panic disorder, and they did not connect any of Roby’s absences to the panic disorder.
McKesson suspended Roby pending an investigation. The main focus of the investigation was to confirm that the number of Roby’s occasions had been calculated correctly, that nothing in her personnel file excused these occasions, and that the frequency of the occasions justified termination under the attendance policy. During this investigation, Schoener reported that Roby *697had misunderstood her statement about a “new start”; in making the statement, Schoener had meant only that, after the beginning of the new year, Roby would be able to use newly accmed vacation leave to take days off.
On April 14, 2000, McKesson terminated Roby by telephone, and it sent a followup letter on April 17, 2000. On April 24, 2000, Roby submitted a “Request for Action” contesting McKesson’s decision and asserting that her absences were because of her disability. McKesson reaffirmed Roby’s termination on May 10, 2000.
After the termination, Roby was devastated emotionally and financially. She depleted her savings and lost her medical insurance, which led her to forgo necessary treatment. She developed agoraphobia (anxiety in public places) and became suicidal. In 2001, the United States Social Security Administration found Roby to be completely disabled.
B
In 2001, Roby sued employer McKesson and supervisor Schoener. The matter proceeded to trial, with the jury instructions outlining the following theories of recovery: wrongful termination in violation of public policy (against McKesson only); harassment in violation of the California Fair Employment and Housing Act (FEHA) (Gov. Code, § 12940, subd. (j))1 (against McKesson and Schoener); discrimination in violation of the FEHA (§ 12940, subd. (a)) (against McKesson only); and failure to accommodate in violation of the FEHA (§ 12940, subd. (m)) (against McKesson only).
Three of the claims against McKesson — wrongful termination in violation of public policy, discrimination in violation of the FEHA (§ 12940, subd. (a)), and failure to accommodate in violation of the FEHA (§ 12940, subd. (m))— were based, at least in part, on Roby’s termination, and therefore the damages for these causes of action necessarily overlapped. The trial court instructed the jury that if it found liability on more than one of these theories, it should determine the total economic damages resulting from Roby’s termination and insert that total figure as the economic damages for each of the separate theories of recovery. The court emphasized that the jury should not divide the total economic damages into parts and distribute those parts among the separate theories of recovery; the court assured the jury that the court would count the economic damages only one time no matter how many times the jury inserted the same dollar amounts on the special verdict form.2 As to *698noneconomic damages, however, the trial court instructed the jury that damages could vary for each of the three theories of recovery and that the jury should therefore determine the appropriate amount applicable for each theory.3
In closing argument, plaintiff’s counsel told the jury that plaintiff was seeking a total of $1.5 million in noneconomic damages on all causes of action, and no more.
When the jury first reported that it had reached a verdict, an irregularity appeared. The portion of the verdict that the trial court read specified $1.5 million in damages for each of the four damages categories listed on the special verdict form (past and future economic damages, and past and future noneconomic damages) and then specified $1.5 million as the total of these amounts. At that point, the trial court stopped reading the verdict. The court instructed the jury further, and the jury resumed its deliberations. The court’s additional instructions restated what it had explained earlier, again making clear that, in calculating the total judgment, the court would not add together the economic damages for the three termination-related causes of action, but it would add together the noneconomic damages for those same three causes of action.4
The jury found in favor of Roby on all causes of action. Its special verdict stated these damages:
*699Wrongful termination in violation of public policy — McKesson
Economic losses
Past $605,000
Future $706,000
Noneconomic losses
Past $250,000
Future $250,000
Discrimination — McKesson
Economic losses
Past $605,000
Future $706,000
Noneconomic losses
Past $200,000
Future $100,000
Failure to accommodate — McKesson
Economic losses
Past $605,000
Future $706,000
Noneconomic losses
Past $400,000
Future $400,000
Harassment — McKesson
Noneconomic losses
Past $300,000
Future $300,000
Harassment — Karen Schoener
Noneconomic losses
Past $250,000
Future $250,000
The special verdict indicates that the jury followed the trial court’s instructions as to the three termination-related causes of action (wrongful termination, discrimination, and failure to accommodate). For all three of these causes of action, the jury listed the same amounts for economic losses, but it listed different amounts for noneconomic losses.
The trial court rendered judgment of $3,511,000 against McKesson and $500,000 against Schoener. The judgment of $3,511,000 against McKesson was consistent with the court’s jury instructions in that the court counted the *700economic losses for the three termination-related causes of action ($605,000 and $706,000) only once, but the court treated the jury’s findings of noneconomic losses for these same causes of action cumulatively.
In a separate verdict, the jury found punitive damages of $15 million against McKesson and $3,000 against Schoener. The trial court rendered judgment accordingly.
The trial court later denied defendants’ motions for a new trial and for judgment notwithstanding the verdict. But on stipulation of the parties the court reduced the compensatory damages judgment against McKesson by $706,000, resulting in a net judgment of $2,805,000. This adjustment corrected an apparent jury error in the award of economic losses.5
C
Both defendants appealed. The Court of Appeal determined that the awards of noneconomic damages for the three termination-related causes of action (wrongful termination, discrimination, and failure to accommodate) overlapped one another. Accordingly, the court upheld only the highest of these three awards. This led to an $800,000 reduction in the total compensatory damages award against employer McKesson, resulting in a net compensatory damages award of $2,005,000 for Roby.
In addition, the Court of Appeal held that the evidence was insufficient to support the jury’s harassment verdict. The court focused on our statement in Reno v. Baird (1998) 18 Cal.4th 640 [76 Cal.Rptr.2d 499, 957 P.2d 1333] (Reno) that “ ‘commonly necessary personnel management actions ... do not come within the meaning of harassment.’ ” (Id. at pp. 646-647, quoting Janken v. GM Hughes Electronics (1996) 46 Cal.App.4th 55, 64 — 65 [53 Cal.Rptr.2d 741] (Janken).) The Court of Appeal viewed that statement as indicating a sharp distinction that not only placed discrimination and harassment claims into separate legal categories but also barred a plaintiff from using personnel management actions as evidence in support of a harassment *701claim. The Court of Appeal therefore disregarded every act of defendants that could be characterized as personnel management, and, looking only at the remaining evidence, the court found it insufficient to support the jury’s harassment finding.
Having rejected the jury’s harassment finding, the Court of Appeal deducted an additional $600,000 from the award of compensatory damages against employer McKesson, and it vacated both the compensatory and punitive damages awards against supervisor Schoener. This $600,000 reduction (along with the $800,000 already deducted) resulted in a compensatory damages award against McKesson of $1,405,000. With respect to the $15 million award of punitive damages against McKesson, the Court of Appeal concluded that a significant portion of that award was “no doubt strongly influenced” by the jury’s harassment finding (which the court had vacated). The court also noted that the compensatory damages were substantial (even after the reductions) and included “ ‘outrage’ components that are, to a large extent, duplicated by the punitive damage verdict.” After considering employer McKesson’s substantial net worth and Roby’s financial vulnerability, the Court of Appeal concluded that $2 million in punitive damages (approximately 1.42 times the reduced compensatory damages award of $1,405,000) was the federal constitutional maximum for this case.
The Court of Appeal saw no purpose in remanding the matter for a new trial on the question of punitive damages; instead, it ordered the trial court to modify the judgment against employer McKesson to reflect the reduction of compensatory damages from $2,805,000 to $1,405,000 and the reduction of punitive damages from $15 million to $2 million; it then affirmed the judgment as modified. As to the jury’s verdict against supervisor Schoener for harassment, the Court of Appeal ordered the trial court to render judgment in her favor. The court later denied Roby’s petition for a rehearing.
We granted Roby’s petition for review.
H
A. Did the Court of Appeal Err in Concluding That Three of Roby’s Noneconomic Damages Awards, All of Which Were Related to Some Extent to Her Termination, Overlapped One Another?
In her petition for review, Roby asserted that the Court of Appeal erred when it struck, as duplicative, the jury’s $500,000 award of noneconomic damages for wrongful termination in violation of public policy. Roby, however, did not challenge the Court of Appeal’s decision to strike the $300,000 award of noneconomic damages for discrimination in violation of the FEHA.
*702In Tavaglione v. Billings (1993) 4 Cal.4th 1150, 1158-1159 [17 Cal.Rptr.2d 608, 847 P.2d 574], we explained: “Regardless of the nature or number of legal theories advanced by the plaintiff, he is not entitled to more than a single recovery for each distinct item of compensable damage supported by the evidence. [Citation.] Double or duplicative recovery for the same items of damage amounts to overcompensation and is therefore prohibited. [Citation.] [][]... [1] In contrast, where separate items of compensable damage are shown by distinct and independent evidence, the plaintiff is entitled to recover the entire amount of his damages, whether that amount is expressed by the jury in a single verdict or multiple verdicts referring to different claims or legal theories.”
As mentioned earlier, Roby alleged three termination-related causes of action against McKesson. Specifically, the special verdict form asked the jury to render a verdict as to wrongful termination in violation of public policy, discrimination in violation of the FEHA (§ 12940, subd. (a)), and failure to accommodate Roby’s disability in violation of the FEHA (§ 12940, subd. (m)). The central assertion of a claim of wrongful termination in violation of public policy is that the employer’s motives for terminating the employee are so contrary to fundamental norms that the termination inflicted an injury sounding in tort. (See Tameny v. Atlantic Richfield Co. (1980) 27 Cal.3d 167, 176 [164 Cal.Rptr. 839, 610 P.2d 1330] (Tameny).) Therefore, Roby’s wrongful termination claim necessarily focused exclusively on the termination itself. The FEHA discrimination and failure-to-accommodate claims also depended to a large extent on Roby’s termination, but both these claims were broader in scope, covering official employment actions that preceded the termination (such as duty assignments and the various disciplinary warnings that McKesson gave Roby).
Roby does not assert that any employment action that preceded her termination caused her to incur out-of-pocket losses. Therefore, in terms of economic damages, the three termination-related causes of action all overlapped one another, which explains why the trial court told the jury to insert the same amounts in the spaces on the special verdict form in which the jury was asked to state economic damages for these three causes of action. The court also made clear to the jury that it would not “stack” these awards of economic damages; in other words, the court would count the award of economic damages only once when calculating the judgment.
The noneconomic damages for each of these three termination-related causes of action present a more difficult problem, however. The verdict form defined noneconomic damages as including “mental suffering, loss of enjoyment of life, grief, anxiety, humiliation, emotional distress, and fear, anger, worry.” In assigning a monetary value to this emotional injury, the jury might *703have found, with respect to the two termination-related FEHA causes of action (discrimination and failure to accommodate), that a significant portion of the injury occurred before the termination. That is, the jury might have reasonably found that each individual act of discrimination leading up to Roby’s termination inflicted a separate emotional injury, and it might have found likewise with respect to each failure to accommodate her disability.
Recognizing that noneconomic damages might vary as to each of the three termination-related, claims, the trial court instructed the jury to assess noneconomic damages individually for each cause of action. The court then proceeded to calculate the total award by adding together the several individual awards of noneconomic damages. This procedure, however, could only be justified if the awards of noneconomic damages for each of the three termination-related causes of action were all mutually exclusive. If they overlapped in part, then, to the extent of the overlap, adding the awards together had the effect of compensating Roby multiple times for the same injury.
Roby argues that the awards of noneconomic damages for the three termination-related causes of action did not overlap at all. Conversely, McKesson argues that the awards overlapped completely, the smaller two of these awards being included in the largest. On this basis McKesson argues in favor of the Court of Appeal’s decision to affirm only the largest of the three awards ($800,000 for failure to accommodate), while treating the other two ($500,000 for wrongful termination and $300,000 for discrimination) as duplicative. We find it impossible to determine to a reasonable degree of certainty which of these interpretations of the verdict the jury intended.
Roby’s cause of action for wrongful termination in violation of public policy necessarily focused exclusively on the termination, and the jury awarded $500,000 in noneconomic damages for that cause of action. Moreover, the termination was a factual component of all three termination-related causes of action. McKesson therefore argues that each of the three awards of noneconomic damages should logically include this $500,000 in damages flowing from the termination itself plus any additional amount necessary to compensate Roby for injurious employment actions that preceded the termination. On this basis, McKesson asserts that the three awards overlap.
When, however, this logic is applied to the jury’s verdict, an inconsistency appears. The jury awarded only $300,000 in noneconomic damages for the FEHA discrimination cause of action. As a matter of law, it cannot be that the same termination caused $500,000 in noneconomic damages when litigated as a wrongful termination in violation of public policy, but that it caused only $300,000 in noneconomic damages when litigated as an instance of discrimination in violation of the FEHA, and this discrepancy is especially odd as the *704FEHA discrimination claim was, as a legal matter, the broader of the two claims — that is, it covered both the termination itself and events that preceded the termination.
Roby asserts that the jury actually found three “different wrongs,” each of which “caused a different amount of suffering.” In other words, Roby argues that the three noneconomic damages awards were intended to be mutually exclusive, compensating her for different events. Roby, however, concedes that there is no evidence of an act of discrimination that is separate from her failure to accommodate and wrongful termination claims, and on that evidentiary basis she agrees with McKesson that the $300,000 discrimination award was properly struck as duplicative.
But Roby does not explain, with respect to her failure-to-accommodate claim, how the noneconomic damages could be based solely on events that preceded the termination when, as a legal matter, the same claim also encompassed the termination itself. Roby’s assertion would require us to conclude that the termination caused no noneconomic damages when litigated as a failure to accommodate in violation of the FEHA, but that it caused $500,000 in noneconomic damages when litigated as a wrongful termination in violation of public policy. These discrepancies suggest that the jury did not really understand the various categories of damages listed on the special verdict form.
In addition, it seems highly unlikely that the jury found that Roby suffered 60 percent greater emotional injury from events that preceded the termination (the award of $800,000 for failure to accommodate) than from the termination itself (the award of $500,000 for wrongful termination). This finding is especially odd because the evidence showed that, before the termination, Roby was coming to work regularly and coping with a difficult situation reasonably well, whereas after the termination she became agoraphobic, suicidal, and completely disabled for purposes of employment. Of course, we do not set aside a verdict simply because we deem its factual findings to be highly unlikely or odd, but these points further suggest that the jury did not understand the various categories of damages, making any effort to divine its intent as to its ambiguous verdict difficult at best.6
*705“[A]n appellate court will interpret the verdict if it is possible to give a correct interpretation,” but will reverse if the verdict is “hopelessly ambiguous.” (Woodcock v. Fontana Scaffolding & Equip. Co. (1968) 69 Cal.2d 452, 457 [72 Cal.Rptr. 217, 445 P.2d 881] (Woodcock).) Here, no explanation of the verdict is satisfactory. Therefore, we solicited additional letter briefs addressing whether “the jury’s compensatory damages verdicts [are] so ambiguous ... as to require a remand to the trial court for a new trial limited to determining the amount of compensatory and punitive damages.” Roby’s letter brief stated she preferred to concede the question of overlapping noneconomic damages awards rather than face a new trial; McKesson’s letter brief agreed to this proposal, similarly expressing a desire to avoid a new trial. At oral argument, Roby’s counsel confirmed that this continues to be Roby’s preference.
Because the jury’s intent in making its noneconomic damages awards cannot be determined to a reasonable certainty, a remand for a new trial on damages would ordinarily be appropriate. (See Woodcock, supra, 69 Cal.2d at p. 457.) Instead, we will accept Roby’s concession, by which she agreed to withdraw her challenge to this part of the Court of Appeal’s decision. Accordingly, the validity of the Court of Appeal’s conclusion that the three termination-related noneconomic damages awards fully overlapped one another is no longer in dispute.
B. Did the Court of Appeal Err in Allocating Roby’s Evidence Between Her Harassment Claim and Her Discrimination Claim, and Based on That Allocation, Finding Insufficient Evidence to Support the Harassment Verdict?
Roby challenges the Court of Appeal’s conclusion that there was insufficient evidence to support the jury’s harassment verdict. Specifically, she argues that, under the FEHA, the Court of Appeal should not have excluded personnel management actions as evidence in support of her harassment claim. We agree.
In the FEHA, the terms “discriminate” and “harass” appear in separate provisions and define distinct wrongs. (See Reno, supra, 18 Cal.4th at pp. 645-647; see also State Dept. of Health Services v. Superior Court (2003) 31 Cal.4th 1026, 1040 [6 Cal.Rptr.3d 441, 79 P.3d 556].) As relevant here, subdivision (a) of section 12940 makes it “unlawful” (subject to certain exceptions) “[f]or an employer, because of the . . . physical disability, mental disability, [or] medical condition ... of any person ... to discriminate against the person in compensation or in terms, conditions, or privileges of employment.” (Italics added.) Subdivision (j)(l) of the same statute makes it unlawful (again subject to certain exceptions) “[f]or an employer ... or any *706other person, because of . . . physical disability, mental disability, [or] medical condition ... to harass an employee . . . .” (§ 12940, subd. (j)(l), italics added.)
Because the FEHA treats harassment in a separate provision, there is no reason to construe the FEHA’s prohibition against discrimination broadly to include harassment.7 Hence, our case law makes clear that the FEHA’s discrimination provision addresses only explicit changes in the “terms, conditions, or privileges of employment” (§ 12940, subd. (a)); that is, changes involving some official action taken by the employer. (Reno, supra, 18 Cal.4th at pp. 645-647.) In the case of an institutional or corporate employer, the institution or corporation itself must have taken some official action with respect to the employee, such as hiring, firing, failing to promote, adverse job assignment, significant change in compensation or benefits, or official disciplinary action.
By contrast, harassment often does not involve any official exercise of delegated power on behalf of the employer. We explained this point in Reno: “ ‘Courts have employed the concept of delegable authority as a test to distinguish conduct actionable as discrimination from conduct actionable as harassment. We adopt this approach to find that the exercise of personnel management authority properly delegated by an employer to a supervisory employee might result in discrimination, but not in harassment.’ ”8 (Reno, supra, 18 Cal.4th at p. 646, quoting Janken, supra, 46 Cal.App.4th at p. 64, italics added.) Thus, harassment focuses on situations in which the social environment of the workplace becomes intolerable because the harassment (whether verbal, physical, or visual) communicates an offensive message to the harassed employee.
Because a harasser need not exercise delegated power on behalf of the employer to communicate an offensive message, it does not matter for purposes of proving harassment whether the harasser is the president of the *707company or an entry-level clerk, although harassment by a high-level manager of an organization may be more injurious to the victim because of the prestige and authority that the manager enjoys. When the harasser is a supervisor, the employer is strictly liable for the supervisor’s actions. (State Dept. of Health Services v. Superior Court, supra, 31 Cal.4th at pp. 1040-1041.) When the harasser is a nonsupervisory employee, employer liability turns on a showing of negligence (that is, the employer knew or should have known of the harassment and failed to take appropriate corrective action). (§ 12940, subd. (j)(l).)
These distinctions place discrimination and harassment in separate categories in regard to application of the FEHA; as explained above, discrimination refers to bias in the exercise of official actions on behalf of the employer, and harassment refers to bias that is expressed or communicated through interpersonal relations in the workplace. This conclusion is consistent with our analysis of the FEHA in Reno. There, we said: “ ‘[HJarassment consists of conduct outside the scope of necessary job performance, conduct presumably engaged in for personal gratification, because of meanness or bigotry, or for other personal motives. ...[!]■•• [f] •• • [CJommonly necessary personnel management actions ... do not come within the meaning of harassment. . . . These actions may retrospectively be found discriminatory if based on improper motives, but in that event the remedies provided by the FEHA are those for discrimination, not harassment. . . . This significant distinction underlies the differential treatment of harassment and discrimination in the FEHA.’ ”9 (Reno, supra, 18 Cal.4th at pp. 645-647, quoting Janken, supra, 46 Cal.App.4th at pp. 63-65.)
The FEHA’s distinction between discrimination and harassment does not mean that harassment claims are relegated to a lower status. The FEHA does not differentiate in terms of wrongfulness between discrimination and harassment; both are “unlawful employment practice[s]” (§ 12940), and in both cases an aggrieved employee can obtain full compensation for any resulting injury. In addition, we can discern no reason why an employee who is the victim of discrimination based on some official action of the employer cannot also be the victim of harassment by a supervisor for abusive messages that create a hostile working environment, and under the FEHA the employee would have two separate claims of injury.
Our decision in Miller, supra, 36 Cal.4th at pages 460-466, further clarifies the FEHA’s distinction between discrimination and harassment. Although discrimination and harassment are separate wrongs, they are sometimes closely interrelated, and even overlapping, particularly with regard to proof. In Miller, we considered whether evidence of widespread sexual *708favoritism in the workplace could constitute sexual harassment against the nonfavored employees. We concluded that it could, provided that the favoritism was so severe or pervasive as to alter the working conditions. (36 Cal.4th at p. 466.) Significantly, the favoritism at issue in Miller took the form of official employment actions, including promotions and favorable job assignments given to female employees involved in sexual relationships with a particular male supervisor. (Id. at pp. 452-459.) The Miller plaintiffs, however, were not subject to any demands for sexual favors. (Ibid.) In concluding that the plaintiffs had nevertheless stated a prima facie case of harassment in violation of the FEHA, we stated that widespread sexual favoritism could convey a “demeaning message ... to female employees that they are viewed by management as ‘sexual playthings’ or that the way required for women to get ahead in the workplace is to engage in sexual conduct with their supervisors or the management.” (Miller, at p. 451; see also id. at p. 464.) This demeaning message, we held, could give rise to an actionable hostile work environment. (Id. at p. 451.)
Thus, in Miller the immediate source of the plaintiffs’ alleged injuries was the offensive sex-biased message that the supervisor conveyed, not a demotion or an unfavorable job assignment, and therefore the plaintiffs’ cause of action was for harassment, not for discrimination. Nevertheless, official employment actions constituted the evidentiary basis of the harassment cause of action, because the supervisor used those official actions as his means of conveying his offensive message. Our decision in Miller is wholly consistent with Reno, supra, 18 Cal.4th at pages 645-647, because it confirms that harassment is generally concerned with the message conveyed to an employee, and therefore with the social environment of the workplace, whereas discrimination is concerned with explicit changes in the terms or conditions of employment. Miller, however, makes clear that in some cases the hostile message that constitutes the harassment is conveyed through official employment actions, and therefore evidence that would otherwise be associated with a discrimination claim can form the basis of a harassment claim. Moreover, in analyzing the sufficiency of evidence in support of a harassment claim, there is no basis for excluding evidence of biased personnel management actions so long as that evidence is relevant to prove the communication of a hostile message.
Here, Roby’s discrimination claim sought compensation for official employment actions that were motivated by improper bias. These discriminatory actions included not only the termination itself but also official employment actions that preceded the termination, such as the progressive disciplinary warnings and the decision to assign Roby to answer the office telephones during office parties. Roby’s harassment claim, by contrast, sought compensation for hostile social interactions in the workplace that affected the workplace environment because of the offensive message they conveyed to *709Roby. These harassing actions included Schoener’s demeaning comments to Roby about her body odor10 and arm sores, Schoener’s refusal to respond to Roby’s greetings, Schoener’s demeaning facial expressions and gestures toward Roby, and Schoener’s disparate treatment of Roby in handing out small gifts. None of these events can fairly be characterized as an official employment action. None involved Schoener’s exercising the authority that McKesson had delegated to her so as to cause McKesson, in its corporate capacity, to take some action with respect to Roby. Rather, these were events that were unrelated to Schoener’s managerial role, engaged in for her own purposes.
Miller, however, makes clear that some official employment actions done in furtherance of a supervisor’s managerial role can also have a secondary effect of communicating a hostile message. This occurs when the actions establish a widespread pattern of bias. (Miller, supra, 36 Cal.4th at p. 466.) Here, some actions that Schoener took with respect to Roby are best characterized as official employment actions rather than hostile social interactions in the workplace, but they may have contributed to the hostile message that Schoener was expressing to Roby in other, more explicit ways. These would include Schoener’s shunning of Roby during staff meetings, Schoener’s belittling of Roby’s job, and Schoener’s reprimands of Roby in front of Roby’s coworkers. Moreover, acts of discrimination can provide evidentiary support for a harassment claim by establishing discriminatory animus on the part of the manager responsible for the discrimination, thereby permitting the inference that rude comments or behavior by that same manager were similarly motivated by discriminatory animus.
Therefore, discrimination and harassment claims can overlap as an evidentiary matter. The critical inquiry when a court is deciding whether the evidence is sufficient to uphold a verdict finding both discrimination and harassment is whether the evidence indicates violations of both FEHA prohibitions, but nothing prevents a plaintiff from proving these two violations with the same (or overlapping) evidentiary presentations.
Here, the Court of Appeal allocated Roby’s evidence between her discrimination claim and her harassment claim, and on that basis found the evidence *710of harassment insufficient. The court said: “[M]ost of the alleged harassment . . . was conduct that fell within the scope of Schoener’s business and management duties. Acts such as selecting Roby’s job assignments, ignoring her at staff meetings, portraying her job responsibilities in a negative light, or reprimanding her in connection with her performance, cannot be used to support a claim of hostile work environment. While these acts might, if motivated by bias, be the basis for a finding of employer discrimination, they cannot be deemed ‘harassment’ within the meaning of the FEHA.” (Italics added & omitted.)
The Court of Appeal concluded that, after this “business and management” evidence was “sifted out,” there was little evidence that supervisor Schoener’s hostility toward Roby was based on Roby’s disability rather than mere rudeness. The remaining evidence was limited to Schoener’s demeaning comments and gestures, Schoener’s refusal to respond to Roby’s greetings, and Schoener’s failure to give Roby gifts. According to the Court of Appeal, “th[is] evidence showed that Schoener obviously disliked Roby, shunned her, and showed no compassion for her condition,” but it did not establish that Schoener’s rude treatment of Roby was “because of. . . physical disability, mental disability, [or] medical condition.” (§ 12940, subd. (j)(l), italics added.)
In allocating the evidence between Roby’s discrimination and harassment claims and then ignoring the discrimination evidence when analyzing the harassment verdict, the Court of Appeal erred. As discussed above, tire FEHA treats discrimination and harassment as distinct categories, but nothing in the FEHA requires that the evidence in a case be dedicated to one or the other claim but never to both. Here, the evidence is ample to support the jury’s harassment verdict. The evidence included not only Schoener’s rude comments and behavior, which occurred on a daily basis, but also Schoener’s shunning of Roby during weekly staff meetings, Schoener’s belittling of Roby’s job, and Schoener’s reprimands of Roby in front of Roby’s coworkers. This evidence was sufficient to allow the jury to conclude that the hostility was pervasive and effectively changed the conditions of Roby’s employment. (Lyle v. Warner Brothers Television Productions (2006) 38 Cal.4th 264, 278-279 [42 Cal.Rptr.3d 2, 132 P.3d 211].)
Moreover, the jury could infer, based on the discrimination evidence, that supervisor Schoener’s hostility was “because of .. . [Roby’s] medical condition.” (§ 12940, subd. (j)(l), italics added.) Specifically, the jury could draw this inference from the evidence that Schoener — who knew about Roby’s medical condition — applied employer McKesson’s attendance policy without making any accommodation or even inquiring if an accommodation was possible. The jury could also draw this inference from the degrading *711manner in which Schoener would announce to the office that Roby was “absent again” and from the demeaning comments, gestures, and facial expressions Schoener made in response to Roby’s body odor and arm sores. Viewed together, the evidence is sufficient to support the jury’s conclusion that Schoener harassed Roby in violation of the FEHA.
McKesson concedes that the same evidence can be used in support of both a discrimination claim and a harassment claim. But, citing our decision in Reno, supra, 18 Cal.4th at pages 645-647, McKesson asserts that “nonabusive actions by a supervisor acting in the course of his or her managerial duties” may not support a harassment claim. Whether or not McKesson accurately describes the law, discrimination is by its nature an abusive action, not a “nonabusive action.” Therefore, from the evidence that Schoener discriminated against Roby based on Roby’s medical condition, the jury could reasonably infer that Schoener’s constant hostility toward Roby was also based on her medical condition, thus constituting harassment in violation of the FEHA. (§ 12940, subd. (j)(l).)
It is appropriate, therefore, to reinstate the jury’s harassment verdict against employer McKesson and supervisor Schoener, and it is also appropriate to reinstate the jury’s $3,000 punitive damages award against supervisor Schoener. This conclusion, however, raises two issues that the Court of Appeal did not reach. First, is the $600,000 harassment award against McKesson based in large part on McKesson’s vicarious liability for the harassing acts of its supervisor (see State Dept. of Health Services v. Superior Court, supra, 31 Cal.4th at pp. 1040-1041), and does it therefore duplicate the $500,000 harassment award against Schoener? Second, assuming that the $600,000 award against McKesson includes as its major component McKesson’s vicarious liability for the $500,000 award against Schoener, what evidence if any justifies the additional $100,000 in harassment damages that the jury awarded against McKesson? At oral argument, Roby’s counsel said that, to avoid a remand to the Court of Appeal, Roby would stipulate to a lower award of $500,000 against McKesson, and McKesson’s counsel accepted this proposed solution of the issue. In other words, Roby conceded that the two harassment awards fully overlapped one another, and that there was insufficient evidence to support a harassment award against McKesson that was independent of the award against Schoener. Accordingly, we will direct the Court of Appeal to modify the trial court’s judgment to provide for a single harassment award of $500,000 against both McKesson and Schoener. (See § 12940, subd. (j)(l), (3); see also State Dept. of Health Services v. Superior Court, supra, 31 Cal.4th at pp. 1040-1041.)
*712C. Did the Jury’s Award of Punitive Damages Against McKesson Exceed the Amount Permitted Under the Federal Constitution?
In a civil case not arising from the breach of a contractual obligation, the jury may award punitive damages “where it is proven by clear and convincing evidence that the defendant has been guilty of oppression, fraud, or malice.” (Civ. Code, § 3294, subd. (a).) “Malice” is defined as intentional injury or “despicable conduct which is carried on by the defendant with a willful and conscious disregard of the rights or safety of others.” (Id., § 3294, subd. (c)(1).) “Oppression” is defined as “despicable conduct that subjects a person to cruel and unjust hardship in conscious disregard of that person’s rights.” (Id., § 3294, subd. (c)(2).) Employer McKesson did not petition this court for review of the Court of Appeal’s decision, and therefore it has effectively conceded that the evidence here supports an award against it of punitive damages. The question remains, however, whether the $15 million award against McKesson is consistent with federal constitutional constraints. The Court of Appeal held that in this case $2 million marked the uppermost constitutional limit for punitive damages. Roby asserts on review that the jury’s entire $15 million award falls within the constitutional limit and therefore should be reinstated. We agree with the Court of Appeal that the $15 million award exceeds the federal constitutional limit, but we disagree with the Court of Appeal that in this case the appropriate limit is $2 million.
The due process clause of the Fourteenth Amendment to the United States Constitution places constraints on state court awards of punitive damages. (See State Farm Mut. Automobile Ins. Co. v. Campbell (2003) 538 U.S. 408, 416-418 [155 L.Ed.2d 585, 123 S.Ct. 1513] (State Farm); BMW of North America, Inc. v. Gore (1996) 517 U.S. 559, 568 [134 L.Ed.2d 809, 116 S.Ct. 1589] (BMW).) We recently explained the basis of these constraints: “The imposition of ‘grossly excessive or arbitrary’ awards is constitutionally prohibited, for due process entitles a tortfeasor to ‘ “fair notice not only of the conduct that will subject him to punishment, but also of the severity of the penalty that a State may impose.” ’ [Citation.]” (Simon v. San Paolo U.S. Holding Co., Inc. (2005) 35 Cal.4th 1159, 1171 [29 Cal.Rptr.3d 379, 113 P.3d 63] (Simon).)
In State Farm, the high court articulated “three guideposts” for courts reviewing punitive damages: “(1) the degree of reprehensibility of the defendant’s misconduct; (2) the disparity between the actual or potential harm suffered by the plaintiff and the punitive damages award; and (3) the difference between the punitive damages awarded by the jury and the civil penalties authorized or imposed in comparable cases.” (State Farm, supra, 538 U.S. at p. 418; see also BMW, supra, 517 U.S. at p. 575.) We discuss each below.
*7131. Degree of reprehensibility
Of the three guideposts that the high court outlined in State Farm, supra, 538 U.S. at page 418, the most important is the degree of reprehensibility of the defendant’s conduct. On this question, the high court instructed courts to consider whether “[1] the harm caused was physical as opposed to economic; [2] the tortious conduct evinced an indifference to or a reckless disregard of the health or safety of others; [3] the target of the conduct had financial vulnerability; [4] the conduct involved repeated actions or was an isolated incident; and [5] the harm was the result of intentional malice, trickery, or deceit, or mere accident.” (Id. at p. 419.)
With respect to the first of these reprehensibility factors, the harm to Roby was “physical” in the sense that it affected her emotional and mental health, rather than being a purely economic harm. (State Farm, supra, 538 U.S. at p. 419.) With respect to the second reprehensibility factor, it was objectively reasonable to assume that employer McKesson’s acts of discrimination and harassment toward Roby would affect her emotional well-being, and therefore McKesson’s “conduct evinced an indifference to or a reckless disregard of the health or safety of others.” (Ibid.) The third reprehensibility factor is likewise present here: Roby was a relatively low-level employee who quickly depleted her savings and lost her medical insurance as a result of her termination, and therefore it appears that she “had financial vulnerability.” (Ibid.)
The fourth reprehensibility factor of the high court’s State Farm test, however, is not present here. Supervisor Schoener’s wrongful conduct was certainly repeated, as she subjected Roby to a series of discriminatory disciplinary actions and harassed Roby on an almost daily basis. But there is no indication of repeated wrongdoing by employer McKesson, as discussed below.
With respect to the discrimination claim, employer McKesson’s wrongdoing was limited to its one-time decision to adopt a strict attendance policy that, in requiring 24-hour advance notice before an absence, did not reasonably accommodate employees who had disabilities or medical conditions that might require several unexpected absences in close succession. McKesson’s act of discharging Roby (including the perfunctory investigation that accompanied it) was simply an application of this attendance policy in accordance with its terms. The jury found that McKesson’s adoption of this flawed attendance policy constituted “oppression” or “malice,” justifying an award of *714punitive damages.11 (Civ. Code, § 3294, subd. (a).) Nevertheless, McKesson’s adoption of this attendance policy was a single corporate decision.
With respect to the harassment claim, McKesson’s corporate wrongdoing was also a single event. In considering this issue, it is important to keep in mind that a corporate defendant cannot be punished for harassment merely because one of its employees has harassed another employee in the workplace; rather, the focus of the punitive damages inquiry must be on the corporation’s institutional responsibility, if any, for that harassment. This principle is codified in Civil Code section 3294, subdivision (b), which provides: “An employer shall not be liable for [punitive] damages . . . , based upon acts of an employee of the employer, unless the employer had advance knowledge of the unfitness of the employee and employed him or her with a conscious disregard of the rights or safety of others or authorized or ratified the wrongful conduct.... With respect to a corporate employer, the advance knowledge and conscious disregard, authorization, [or] ratification . . . must be on the part of an officer, director, or managing agent of the corporation.” (Italics added.) In White v. Ultramar, Inc. (1999) 21 Cal.4th 563 [88 Cal.Rptr.2d 19, 981 P.2d 944] (White), we construed the latter statement as requiring the officer, director, or managing agent to be someone who “exercise^] substantial discretionary authority over decisions that ultimately determine corporate policy.” (Id. at p. 577.)
In this case, the Court of Appeal concluded that the jury could reasonably have found supervisor Schoener to be a “managing agent” of employer McKesson. On that basis, the court concluded that the jury’s award of punitive damages could be justified based on Schooner’s actions alone, regardless of whether more senior managers at McKesson were informed of Schoener’s actions. We disagree.
At the time of Roby’s termination, McKesson had over 20,000 employees; Schoener worked at a local distribution center supervising four of them. When we spoke in White about persons having “discretionary authority *715over . . . corporate policy” (White, supra, 21 Cal.4th at p. 577), we were referring to formal policies that affect a substantial portion of the company and that are the type likely to come to the attention of corporate leadership. It is this sort of broad authority that justifies punishing an entire company for an otherwise isolated act of oppression, fraud, or malice. The record here does not support the conclusion that Schoener exercised that sort of broad authority or that she was a “managing agent” for purposes of awarding punitive damages under Civil Code section 3294, subdivision (b). Therefore, in assessing the reprehensibility of employer McKesson’s conduct, we must look to what McKesson’s more senior managers knew and did.
The record only weakly supports the jury’s finding that a “managing agent” of employer McKesson was informed of Schooner’s unlawful harassment of Roby and ratified it, either expressly or by inaction. It is true that Roby complained more than once to the manager of her distribution center about her ongoing conflicts with Schoener, but personality clashes in the workplace are not uncommon, and Roby’s complaints did not link these conflicts to her medical condition and therefore did not put McKesson on specific notice that Schoener was violating Roby’s FEHA rights. Nevertheless, the evidence indicates that Roby once met with two mid-level managers (the head of Roby’s distribution center and the regional human resources director) and told them of Schoener’s ongoing harassment, expressly linking that harassment to her medical condition. Roby testified as follows about this meeting: “I told them that, yes; that I was being harassed once again by . . . Schoener .... She had made derogatory remarks that day that was upsetting, and it was public, fit] ... fit] .. . [I]t had to do with the head sweats that I had, and I was digging at my arms again.”
McKesson does not argue that the mid-level managers at this meeting were not “managing agent[s]” for purposes of awarding punitive damages under Civil Code section 3294, subdivision (b), and therefore we will assume for purposes of this appeal that at least one of them was a managing agent. Hence, Roby’s statement at this meeting, combined with the more general complaints that Roby made, constituted sufficient evidence to support the jury’s inference that a McKesson managing agent eventually became aware of Schoener’s unlawful harassment of Roby. That McKesson thereafter continued to employ Schoener as Roby’s supervisor without taking any corrective measures indicates “conscious disregard of the rights or safety of others” (Civ. Code, § 3294, subd. (b)), thus warranting punitive damages.
Nevertheless, the evidence establishing corporate wrongdoing in regard to supervisor Schoener’s unlawful harassment of Roby does not indicate any repeated corporate misconduct. There is no evidence, for example, that Schoener’s actions toward Roby were the product of a corporate culture that *716encouraged similar supervisorial conduct. Rather, they appear to be the isolated actions of a single supervisor, combined with the one-time failure on the part of employer McKesson to take prompt responsive action when these events came to its attention.
With respect to the fifth reprehensibility factor listed in the high court’s State Farm decision, in awarding punitive damages against McKesson the jury here necessarily determined that McKesson acted with “conscious disregard” of the rights of others (Civ. Code, § 3294, subd. (c)(1), (2)); therefore, the conduct at issue was certainly not “mere accident” (State Farm, supra, 538 U.S. at p. 419). Nevertheless, the corporate conduct falls short of “intentional malice.” (Ibid.) The evidence does not suggest that employer McKesson adopted the attendance policy in question — and in particular the requirement of 24-hour advance notice for all absences — with a purpose or motive to discriminate. Rather, McKesson’s apparent purpose in requiring 24-hour advance notice was to enable advance planning by its supervisors and thus ensure adequate staffing levels on a daily basis. McKesson’s wrongdoing was more a failure to prevent the foreseeable discriminatory consequences flowing from its otherwise appropriate attendance policy than it was an act rooted in “intentional malice.”
We focus here on McKesson’s adoption of the attendance policy and not on the conduct of McKesson’s mid-level managers who applied the policy in reviewing Roby’s grievance and determining to uphold her termination. For reasons stated above (see p. 715, ante), we will assume for purposes of this appeal that at least one of these mid-level managers was a “managing agent” under Civil Code section 3294, subdivision (b). But there is no evidence that they were empowered, when reviewing Roby’s grievance, to make an on-the-spot accommodation in abrogation of the terms of McKesson’s attendance policy. To the contrary, the evidence indicated that they were required to enforce the policy strictly. At most, they could have retroactively reclassified some of Roby’s occasions as protected medical leave under the FMLA (29 U.S.C. § 2601 et seq.), but Roby failed to submit adequate documentation to support such a reclassification, even after being told that this was necessary.
We need not decide whether McKesson’s managers were required to do more than they did to assist Roby in establishing FMLA eligibility, because in any case their conduct was not so “despicable” (Civ. Code, § 3294, subd. (c)) as to support a finding that they acted with “oppression, fraud, or malice” (id., § 3294, subd. (a)), warranting an award of punitive damages. Roby had missed work without notice 11 times in a period of about 15 months, and these abrupt absences had continued despite progressive disciplinary warnings. During these months, Roby had never asked that her absences be treated as FMLA leave, although she had taken FMLA leave for other absences. In *717addition, although Roby’s supervisors were generally aware of her panic attacks, Roby’s own understanding of her medical condition evolved over time, and therefore her reports about this condition to her supervisors lacked specificity regarding the accommodations she might need. She never submitted a medical report relating her absences to her panic disorder, and the only medical documents in her personnel file that even mentioned the panic disorder stated that it was “not contagious” and that it was “stabilized” with medication. These brief medical notes nowhere suggested that the panic disorder interfered with Roby’s ability to work or constituted a “ ‘serious health condition’ ” (29 U.S.C. § 2611(11)) justifying FMLA leave. For these reasons, the conduct of the mid-level managers who reviewed and approved Roby’s termination does not provide an independent basis for awarding punitive damages against McKesson.
In regard to employer McKesson’s failure to take responsive action once it learned of supervisor Schoener’s unlawful harassment of Roby, we again see no indication of a corporate purpose to cause injury to Roby. Rather, McKesson’s failure to take appropriate action is better characterized as managerial malfeasance. This failure is not excusable, but it is partly explainable by the somewhat vague nature of Roby’s complaints. As noted earlier, the record indicates only a single instance when Roby’s complaint to mid-level managers linked the ongoing harassment to a medical condition. This complaint should have alerted McKesson to respond, and hence the jury’s punitive damages award against McKesson finds sufficient support in the evidence. But McKesson’s conduct, although wrongful, does not rise to the kind of oppressive, fraudulent, or malicious conduct that has in the past justified large punitive damages awards. (See, e.g., Romo v. Ford Motor Co. (2003) 113 Cal.App.4th 738 [6 Cal.Rptr.3d 793] [the defendant mass-produced and sold a vehicle it knew to be designed in a way that was inherently dangerous to human life; three people died; three others were injured; punitive damages: $23,723,287]; Rufo v. Simpson (2001) 86 Cal.App.4th 573 [103 Cal.Rptr.2d 492] [the defendant maliciously stabbed and killed two people; punitive damages: $25 million]; Weeks v. Baker & McKenzie (1998) 63 Cal.App.4th 1128 [74 Cal.Rptr.2d 510] [a partner of the defendant law firm put his hand in the breast pocket of his secretary’s blouse, made a grabbing gesture toward her breasts, touched her buttocks, and made sexually harassing statements; the defendant law firm was aware of numerous prior incidents of severe sexual harassment involving the same partner; punitive damages: $3.5 million].)
Taking into account all five reprehensibility factors that the high court set forth in State Farm, supra, 538 U.S. at page 419, we conclude that employer McKesson acted wrongfully and in a manner warranting civil penalties; nevertheless, the reprehensibility of McKesson’s conduct was at the low end of the range of wrongdoing that can support an award of punitive damages *718under California law, notwithstanding the seriousness of Roby’s emotional injury and her financial vulnerability.
2. Disparity between actual harm and punitive damages
The second guidepost that the United States Supreme Court articulated in State Farm for assessing the constitutionality of a punitive damages award is “the disparity between the actual or potential harm suffered by the plaintiff and the punitive damages award.” (State Farm, supra, 538 U.S. at p. 418.) Here, the trial court reduced the jury’s award of $3,511,000 in compensatory damages against employer McKesson to $2,805,000. The Court of Appeal further reduced this figure to $1,405,000. But our conclusion in part IIB, ante, requires reinstatement of the jury’s $500,000 harassment award against supervisor Schoener, for which employer McKesson is also liable (see p. 711, ante), resulting in a total compensatory damages award of $1,905,000. Only $605,000 of this sum was for Roby’s economic losses; the remaining $1.3 million in compensatory damages was awarded solely for Roby’s physical and emotional distress and may have reflected the jury’s indignation at McKesson’s conduct, thus including a punitive component. Pertinent here is this statement from our decision in Simon, supra, 35 Cal.4th at page 1189: “[D]ue process permits a higher ratio between punitive damages and a small compensatory award for purely economic damages containing no punitive element than [it does] between punitive damages and a substantial compensatory award for emotional distress; the latter may be based in part on indignation at the defendant’s act and may be so large as to serve, itself, as a deterrent.”
In State Farm, the high court suggested that a ratio of one to one might be the federal constitutional maximum in a case involving, as here, relatively low reprehensibility and a substantial award of noneconomic damages: “When compensatory damages are substantial, then a lesser ratio, perhaps only equal to compensatory damages, can reach the outermost limit of the due process guarantee.” (State Farm, supra, 538 U.S. at p. 425, italics added.)
3. Civil penalties authorized in comparable cases
Finally, we consider “the difference between the punitive damages awarded by the jury and the civil penalties authorized or imposed in comparable cases,” the last of the three guideposts the high court set forth in State Farm, supra, 538 U.S. at page 418, to assess the constitutionality of punitive damages awards. If in this case Roby had pursued her FEHA claims administratively before California’s Fair Employment and Housing Commission, the commission could have assessed a fine against employer McKesson in *719addition to awarding compensatory damages. (§ 12970, subds. (a), (c), (d).) This administrative fine cannot exceed $150,000 (§ 12970, subd. (a)(3)), which of course is tiny by comparison to the jury’s punitive damages award here of $15 million against employer McKesson. Obviously, this guidepost weighs in favor of a lower constitutional limit in this case.
4. Summary
After applying the test that the high court articulated in State Farm, supra, 538 U.S. at page 418, we conclude that a one-to-one ratio between compensatory and punitive damages is the federal constitutional limit here. We base this conclusion on the specific facts of this case. We note in particular the relatively low degree of reprehensibility on the part of employer McKesson and the substantial compensatory damages verdict, which included a substantial award of noneconomic damages.
The concurring and dissenting opinion asserts that a higher ratio— two to one — is appropriate here because of McKesson’s wealth, among other things. (See conc. & dis. opn. of Werdegar, J., post, at p. 723.) It is certainly relevant for a reviewing court to consider the wealth of a defendant when applying federal constitutional limits to an award of punitive damages, thereby ensuring that the award has the appropriate deterrent effect, but the punitive damages award must not punish the defendant simply for being wealthy. (Simon, supra, 35 Cal.4th at pp. 1185-1186.) As the high court said in State Farm, wealth “ ‘provides an open-ended basis for inflating awards’ ” (State Farm, supra, 538 U.S. at pp. 427-428, quoting BMW, supra, 517 U.S. at p. 591 (conc. opn. of Breyer, J.)) and “cannot justify an otherwise unconstitutional punitive damages award” (id. at p. 427). In applying the federal Constitution here, we have taken McKesson’s wealth into consideration, and more to the point we have taken into consideration the deterrent effect that is appropriate in light of McKesson’s wrongdoing. We nevertheless conclude that punitive damages in an amount equal to compensatory damages marks the constitutional limit in this case and still provides the appropriate deterrence. The concurring and dissenting opinion concedes that the jury’s award of $15 million in punitive damages against McKesson far exceeds what the federal Constitution permits. (See conc. & dis. opn. of Werdegar, J., post, at p. 720.) The only disagreement is whether the constitutional limit in this case is equal to the compensatory damages award of $1,905,000, as we hold, or whether it is double that amount, as the concurring and dissenting opinion contends. Based on the relatively low degree of reprehensibility and the substantial award of noneconomic damages, we conclude that $1,905,000 is the maximum punitive damages that may be awarded against employer McKesson in this case in light of the constraints imposed by the federal *720Constitution. Instead of ordering a retrial on the question of punitive damages, we simply direct a reduction of those damages to the $1,905,000 maximum. (See Simon, supra, 35 Cal.4th at pp. 1187-1188.)
Disposition
The Court of Appeal’s judgment is reversed, and the matter is remanded to that court with directions (1) to reinstate a single harassment award of $500,000 against both employer McKesson and supervisor Schoener; (2) to reinstate the jury’s $3,000 punitive damages award against supervisor Schoener; and (3) to modify the punitive damages award against employer McKesson to $1,905,000. The Court of Appeal is directed to affirm the judgment of the trial court as so modified.
George, C. J., Baxter, J., Chin, J., and Corrigan, J., concurred.
All further statutory references are to the Government Code unless otherwise indicated.
Specifically, the trial court instructed the jury: “If you find the defendant liable on two of the verdicts, for example, and you are then moving onto the dollar amounts here, the economic *698loss will be identical for those two. H] . . . [Y]ou would put the same number there, past economic loss and future economic loss, on those lines [of the verdict form]. H]... H] ... [S]o you’re saying, [for example] well, we think that she ought to get a guilder for economic loss. [][]... [B]ut we’re finding it on different ones, because we’re going to put half in this verdict, and half in the other. Don’t do that. You figure out what it is. [f] You put it in there .... [W]hatever your amount is, whatever the dollars are. H] . . . If you bring back a verdict that has dollar amounts in two separate parts of the verdict form, I will know that I only can order judgment for that amount[] [o]nce[.] It’s not going to be that, oh it’s one guilder for one and two and three, and I’m going to stack.”
The trial court stated: “When it comes to non-economic loss . . . , this is a little different, [ft] You may find that the defendants have acted wrongly, and that there’s been . . . non-economic loss. And so you will say, well, for that particular wrongdoing, how much is that worth? [f] Then you’ll come up with an amount, if any. Then you get to the next one. You say, oh, we find that there’s wrong conduct here as well. [][]... That had a different amount. . . caused a different amount of non-economic loss ....[$] For example, one of these verdict forms is for wrongful] discharge; another one is for disparate treatment in the workplace. . . . [][]... [S]o you would put in for that particular one what you think the amount is.”
The trial court stated: “Now, with the economic loss . . . , as I said, if you’ve picked one number in one of the verdict forms, then that number will just transfer for that same line [on] all the other verdict forms. That’s not necessarily the case for the mental suffering, loss of. . . enjoyment of life, and a jury could possibly find that while, for example, starting with wrongful discharge [in] violation of public policy, that led to a certain amount of mental suffering, and it’s worth a certain amount of money, that when you . . . then go to the next verdict form, you can say that caused a different amount of suffering and that’s worth a *699different amount of money . ... ffl . [I]f you think ... it was a different amount [for] each, then you need to figure out how much goes with one claim, how much goes with another, that sort of thing . . . .”
The amounts the jury listed for past and future economic losses ($605,000 and $706,000, respectively) corresponded almost exactly to amounts that Roby’s expert witness had mentioned in his testimony, but in selecting these figures the jury apparently misunderstood the expert’s use of the terms “future losses” and “present value.” Roby’s expert witness had testified that Roby’s total past and future losses in income were $706,299, and he determined the present value of this amount to be $604,657. The jury apparently rounded off the $604,657 figure as the basis for its award of $605,000 for “past economic loss,” and it apparently rounded off the $706,299 figure as the basis for its award of $706,000 for “future economic loss,” resulting in a total economic loss of $1,311,000. In short, the jury duplicated the economic losses by adding the present value of the losses to the aggregate future amount of the same losses.
Other problems — already noted — give us additional concern about the jury’s understanding of the various damages categories. First, the jury initially awarded $1.5 million in damages for each of the damages categories listed on the special verdict form and also specified $1.5 million as the total of these amounts. (See, ante, at p. 698.) It is, of course, highly unlikely that the termination would result in exactly the same damages for each of the damages categories. Second, the jury apparently misunderstood the terms “future losses” and “present value" and therefore duplicated the economic losses by adding the present value of the losses to the aggregate future amount of the same losses. (See, ante, at p. 700, fn. 5.)
Title VII of the federal Civil Rights Act of 1964 (42 U.S.C. § 2000e et seq.) (hereafter Title VE) has no express provision addressing workplace harassment, but courts have construed Title VE’s prohibition against discrimination to include harassment that is sufficiently severe or pervasive to alter the conditions of employment. (See Meritor Savings Bank v. Vinson (1986) 477 U.S. 57, 64-67 [91 L.Ed.2d 49, 106 S.Ct. 2399].)
Notwithstanding this statement in Reno, we have in the past categorized quid pro quo sexual harassment (in which a job benefit is conditioned upon sexual favors and therefore an actual or potential exercise of delegated authority is at issue) as a type of harassment. (See, e.g., Hughes v. Pair (2009) 46 Cal.4th 1035, 1043 [95 Cal.Rptr.3d 636, 209 P.3d 963]; Miller v. Department of Corrections (2005) 36 Cal.4th 446, 461 [30 Cal.Rptr.3d 797, 115 P.3d 77] (Miller).) This conclusion follows the California Code of Regulations, which defines sexual harassment as including “[s]exual favors, e.g., unwanted sexual advances which condition an employment benefit upon an exchange of sexual favors.” (Cal. Code Regs., tit. 2, § 7287.6, subd. (b)(1)(D).)
See page 706, footnote 8, ante.
The Court of Appeal suggested that supervisor Schoener’s demeaning comments about Roby’s body odor were necessary personnel management actions, not acts of harassment, because Schoener needed to take action in response to the complaints of other employees. (See Hannoon v. Fawn Eng’g Corp. (8th Cir. 2003) 324 F.3d 1041, 1047 [Title VII case].) Here, however, the evidence supports the jury’s conclusion that Schoener handled the matter in a way that was unnecessarily demeaning, including reprimanding Roby in front of coworkers and telling Roby “to take more showers.” It was the demeaning manner in which Schoener addressed this issue that constituted the harassment.
We do not mean to suggest that in all FEHA discrimination cases involving attendance policies like the one here at issue, an award of punitive damages will always be supportable based on the employer’s mere adoption of such a policy. It was the application of McKesson’s rigid attendance policy to terminate Roby that ultimately gave rise to McKesson’s liability for her wrongful discharge and the related punitive damages, not the mere adoption of the policy itself. But with regard to the further assessment, under State Farm, of the degree of reprehensiblity of McKesson’s conduct for the specific purpose of determining the maximum constitutionally allowable award of punitive damages in this case, a broader forcus is appropriate. Because the midlevel managers who applied the strict attendance policy to terminate Roby lacked discretion to deviate from it under its very terms, we find the adoption of the policy itself most relevant to an assessment of the overall degree of reprehensibility of McKesson’s misconduct.