In Harnish v. Children’s Hosp. Medical Center, 387 Mass. 152, 155 (1982), we said that “a physician owes to his patient the duty to disclose in a reasonable manner all significant medical information that the physician possesses or reasonably should possess that is material to an intelligent decision by the patient whether to undergo a proposed procedure.” We defined “materiality” as “the significance a reasonable person, in what the physician knows or should know is his patient’s position, would attach to the disclosed risk or risks in deciding whether to submit or not to submit to surgery or treatment.” Id. at 156, quoting from Wilkinson v. Vesey, 110 R.I. 606, 627 (1972). In this case, we consider whether the evidence presented at trial, viewed most favorably to the plaintiffs, was legally sufficient to warrant a finding that the defendant physician violated that duty. More precisely, the question is whether the jury properly could have found that the physician failed to disclose to the patient medical information that the physician reasonably should have recognized as material to the patient’s decision whether to undergo surgery. See Harnish, supra at 155-156.2
The plaintiffs commenced this action in April, 1980, alleging that the defendant, Albert R. Frederick, Jr., a physician, negligently prescribed for the plaintiff Wilfred Precourt a drug called Prednisone, and that, as a result, Precourt developed severe damage to the bones of both his hips. Precourt sought damages for his personal injuries, and his wife, Elizabeth, sought damages for her loss of consortium. After our decision in Harnish, supra, the plaintiffs amended their complaint to allege, in addition to negligence, that, although Frederick knew or reasonably should have known that the use of Prednisone presented a risk of the type of hip damage that Precourt sus*691tained, Frederick nevertheless prescribed Prednisone for Precourt without informing him of that risk.3 The trial judge denied Frederick’s motions for directed verdicts, but the jury found for Frederick on the counts alleging negligence. The jury found for Precourt on his lack of informed consent claim, and they awarded Precourt $800,000 and Elizabeth $200,000. Frederick moved for judgment notwithstanding the verdicts or for a new trial. The judge denied Frederick’s motion, although he conditioned his denial of a new trial of Elizabeth’s claim on her remitting $100,000, which she thereafter did. Frederick appealed from the judgments for the plaintiffs and from the denial of his motion for judgment notwithstanding the verdict, and we transferred the case to this court on our own motion. Frederick challenges the denial of his motions and he challenges various aspects of the judge’s charge to the jury as well. We do not reach the latter challenges because we hold that the evidence was legally insufficient to warrant submission of the case to the jury. We reverse the judgments for the plaintiffs.
We summarize the evidence. In doing so, we follow the familiar rule that, “in reviewing the denial of the defendant’s motions for directed verdict and judgment notwithstanding the verdict, we will construe the evidence most favorably to the plaintiff and disregard that favorable to the defendant.” Cimino v. Milford Keg, Inc., 385 Mass. 323, 326 (1982). In April, 1976, while Precourt was repairing a truck in his yard, some metal flew into his left eye and became embedded there. A physician removed a sliver from just beneath the eye’s surface, but Precourt’s eye failed to improve. By August, 1976, the eye was almost legally blind and was sensitive to light. The physician discovered a piece of metal lodged in the back of Precourt’s eye, and he referred Precourt to Frederick, an ophthal*692mic surgeon practicing at the Massachusetts Eye and Ear Infirmary. Frederick specialized in surgery involving the backs of human eyes. After examining Precourt’s eye, Frederick advised him to undergo surgery to remove the metal. The proposed surgery had a ninety per cent probability of restoring vision to the eye.
In September, 1976, Precourt was admitted to the Massachusetts Eye and Ear Infirmary, and Frederick surgically removed the piece of metal from the retina of Precourt’s eye. After surgery, Frederick placed Precourt on a course of treatment with the drug Prednisone, a steroid used by ophthalmic surgeons to control inflammation. Precourt took Prednisone for approximately ninety days following the operation. Initially, the vision in Precourt’s left eye improved, but thereafter it deteriorated so that, by December, 1976, that eye was blind. Nevertheless, Precourt returned to work as an electrician.
In the spring of 1977, Precourt and Frederick discussed the possibility of a second operation to remove scar tissue that had resulted from the first operation. Frederick told Precourt that a second operation had only a ten per cent probability of restoring vision to the eye. In March, 1977, Precourt again entered Massachusetts Eye and Ear Infirmary, and Frederick performed a second operation on Precourt’s left eye. After that operation, Frederick placed Precourt on Prednisone for approximately fifty-five more days. Precourt never regained vision in his eye.
In the succeeding years, Precourt’s hips bothered him, and in March, 1980, physicians diagnosed his problem as aseptic necrosis of both hips. Aseptic necrosis involves the death of the bones of the joint. Prednisone caused Precourt’s aseptic necrosis. At the time of trial, Precourt had undergone a total replacement of his right hip, and he planned to undergo the same procedure on his left hip. As a result of his hip problems, Precourt could no longer work as an electrician.
The evidence also showed the following. Frederick prescribed Prednisone after surgery for fifty to seventy-five per cent of the patients on whom he operated. Frederick had practiced ophthamology since 1963, and he had not known of any of his patients’ developing aseptic necrosis following the use of *693Prednisone. However, before operating on Precourt, Frederick knew from reading, attending conferences and meetings, and discussions with colleagues, of an association between the use of Prednisone and aseptic necrosis. Frederick knew that, when aseptic necrosis develops, the process of bone death is irreversible, and he described aseptic necrosis as one of the “most prominent” musculoskeletal complications of Prednisone. Frederick also knew that the Physician’s Desk Reference — a book frequently referred to by Frederick and other physicians — listed forty-one possible complications associated with the use of Prednisone, including aseptic necrosis, and that it listed hypertension (elevated blood pressure) and renal impairment as relative contraindications for Prednisone use. Frederick knew before operating on Precourt that Precourt drank six “beers” daily, that he had slightly elevated blood pressure, and that he had a history of gouty arthritis and of a kidney stone.
Precourt testified that Frederick did not mention to him either the likelihood that he would have to take Prednisone or the reported association between Prednisone and aseptic necrosis. Precourt testified that after the first operation Frederick mentioned Prednisone only to prescribe it and to explain its purpose. Precourt also testified that, before the second operation, Frederick told him that he “had everything to gain and nothing to lose” from the proposed surgery. In his testimony, Frederick agreed that he did not tell Precourt that “he was in danger of possibly coming down with aseptic necrosis of the hip.”
Barry Fisherman, an ophthalmologist called as a witness by the plaintiffs, testified that, in his opinion, before prescribing Prednisone for a patient a physician should inform the patient of the major risks of Prednisone use. He also testified that aseptic necrosis was one of those risks. In response to a hypothetical question, Fisherman testified that, because Precourt had “hypertension and arthritis and the history of alcohol consumption as well as kidney stones,” Precourt should have been informed that the Prednisone “could cause additional complications to those systems, such as the hypertension could be worse. There could be a development of a peptic ulcer which could perforate and bleed and the musculoskeletal system, *694such as the arthritis, could be made worse with osteoporosis or [aseptic] necrosis.”
Melvin W. Kramer, an internist called by the plaintiffs, testified that the development of the side effects of Prednisone relates directly to the amount of Prednisone taken and the length of time that the patient takes it, that Precourt’s treatment included a “high dose, long course of therapy,” and that, in his opinion, the cumulative effect of the courses of Prednisone that Precourt took caused Precourt to develop aseptic necrosis. Kramer testified that a physician “is obliged to make warning statements regarding possibilities that the medicine may alter or change” medical conditions such as those contained in Precourt’s history, and that “[t]he combination of sustained steroid therapy in a cumulative fashion adding up to many weeks, many months of therapy, has the ability to alter body chemistry in a known and critical fashion. And patients who have associated medical conditions, such as alcohol consumption, or other such metabolic conditions, the patient then becomes a high risk for the development of certain particular disorders. One of these disorders is aseptic necrosis of the bone or hip.”
The risk that materialized was that Precourt developed aseptic necrosis as a result of the systemic use of Prednisone. Therefore, in order to be entitled to recover damages, the plaintiffs had the burden of proving that Frederick had a duty to inform Precourt in a reasonable manner about that risk. Harnish v. Children’s Hosp. Medical Center, 387 Mass. 152, 157-158 (1982). Frederick had that duty only if he had, or reasonably should have had, information about that risk that he reasonably should have recognized Precourt would consider important in deciding whether to undergo the proposed surgery. Id. at 155-156.
The materiality of information about a potential injury is a function not only of the severity of the injury, but also of the likelihood that it will occur. Regardless of the severity of a potential injury, if the probability that the injury will occur is so small as to be practically nonexistent, then the possibility of that injury occurring cannot be considered a material factor *695in a rational assessment of whether to engage in the activity that exposes one to the potential injury. In Harnish, supra at 156, we recognized the relevancy of the probability factor when we said that appropriate information for disclosure may include “the nature and probability of risks involved.” Similarly, in Rogers v. Commissioner of the Dep’t of Mental Health, 390 Mass. 489, 506 (1983), in discussing substituted judgment decisions about treatment plans for incompetent institutionalized mental patients, we said that a judge must consider, among other factors, the probability of adverse side effects, which “includes an analysis of ‘the severity of these side effects, the probability that they would occur, and the circumstances in which they would be endured. ’ [Guardianship of Roe, 383 Mass. 415, 447 (1981)].”
In order for the jury properly to have determined whether Frederick failed to disclose to Precourt information that Frederick should reasonably have recognized as material to Precourt’s decision, therefore, the jury had to have information about both the severity of aseptic necrosis and the likelihood that it would occur after the use of Prednisone. See Smith v. Shannon, 100 Wash. 2d 26, 33 (1983) (“The determination of materiality is a 2-step process. Initially, the scientific nature of the risk must be ascertained, i.e., the nature of the harm which may result and the probability of its occurrence. . . . The trier of fact must then decide whether that probability of that type of harm is a risk which a reasonable patient would consider in deciding on treatment”); LaCaze v. Collier, 434 So. 2d 1039, 1046 (La. 1983) (“Materiality is, in essence, the product of the risk and its chance of occurring. A severe consequence, ordinarily of interest to the patient, would not require disclosure if the chance of the consequence occurring was so remote as to be negligible. Likewise, no disclosure would be required of a very minor consequence even though the probability of occurrence was high. At trial, once the severity and probability of the risk is presented, the trier of fact may determine materiality without the further aid of expert testimony”); Winkjer v. Herr, 277 N.W.2d 579, 588 (N.D. 1979) (“There is no need to disclose risks of little consequence, those that are *696extremely remote, or those that are common knowledge as inherent in the treatment”); Beauvais v. Notre Dame Hosp., 120 R.I. 271, 276 (1978) (in affirming the granting of a directed verdict for the defendant physician in a case involving the doctrine of informed consent, the court said: “While it is clear that defendant failed to disclose the classic hazards of spinal anesthesia, no evidence regarding the severity or likelihood of those risks was presented so that a jury could determine the materiality of those risks to a reasonable person in plaintiff’s position. See Wilkinson v. Vesey, 110 R.I. at 627-28 . . .”).
In this case, there was no evidence of the likelihood that a person would develop aseptic necrosis after taking Prednisone or that Frederick knew or should have known that the likelihood was other than negligible. Therefore, as a matter of law, the plaintiffs failed to show that Frederick recognized or reasonably should have recognized that the undisclosed risk was material to Precourt’s decision. Characterization of Precourt’s Prednisone dosage as “high” and the course of treatment as “long,” in combination with the evidence that the probability of aseptic necrosis increases as the exposure to Prednisone increases, does not permit the inference that Frederick reasonably should have recognized that the possibility that Precourt would develop aseptic necrosis was material to Precourt’s decision. Nor is such an inference made possible by the evidence of Precourt’s preexisting medical condition or by the evidence that aseptic necrosis is one of the most prominent musculoskeletal complications of Prednisone, or that the risk was “high.” “High” is a relative word. It could mean one in ten, but it could just as well mean one in a million.
The evidence did not warrant a finding that Frederick violated a duty he owed to Precourt. A contrary result is not required by the testimony of the expert witnesses with respect to their essentially legal conclusion that Frederick “should have” made a disclosure that he did not make. Also, no different result is compelled by Precourt’s testimony that before the second operation Frederick told him that he had “everything to gain and nothing to lose.” Precourt could not reasonably have taken that statement literally to mean that the proposed surgery was free of risk.
*697We continue to adhere to the views we expressed in Harnish, supra at 154-156, that self-determination by the patient is an important value worthy of society’s protection, but that there must be a reasonable accommodation between the patient’s right to know, fairness to physicians, and society’s interest that medicine be practiced in this Commonwealth without unrealistic and unnecessary burdens on practitioners. As we said in Harnish, supra at 156, “[t]he obligation to give adequate information does not require the disclosure of all risks of a proposed therapy.” We also observed in Harnish, supra at 155, that “[t]he remotely possible risks of a proposed treatment may be almost without limit,” and we at least implied, as we now hold, that a physician is not required to inform a patient of remote risks. The development of our law concerning the distinction between risks that as a matter of law may be considered remote, and those that may be left to the determination of a fact finder, must await future cases. It is clear, however, that when, as in this case, the evidence does not permit the jury to draw an inference that the physician knew or reasonably should have known that the probability that a particular risk would materialize was other than negligible, the evidence is insufficient to warrant a finding that the physician violated his duty of disclosure.
Judgments reversed.
We acknowledge the brief of amici curiae Massachusetts Medical Society, Massachusetts Society of Eye Physicians and Surgeons, and Massachusetts Eye and Ear Infirmary.
The original complaint alleged that Elizabeth suffered her injury as a result of Frederick’s negligence. The plaintiffs did not amend the complaint to allege that Elizabeth’s injury resulted from Frederick’s failure adequately to inform Precourt. However, the trial judge and the parties have treated the amended complaint as alleging that Elizabeth’s loss of consortium resulted from Frederick’s failure adequately to inform Precourt, and we do so as well.