(dissenting).
I wholeheartedly agree with the majority that an instruction on the permissible infer-encés which support a finding of constructive possession is unnecessary and undesirable.1 However, if one is given, as was done here, the instruction must be a correct statement of the law. Because the instruction given here was not, I respectfully dissent.
The defendant accurately pointed out to the trial court that the constructive possession instruction was an incomplete statement of the law at best and an incorrect statement of the law at worst. Because I believe that the majority’s opinion confirms that this court no longer adheres to the principles set out in State v. Reeves, 209 N.W.2d 18 (Iowa 1973), I begin with a discussion of that case.
In Reeves, we were required for the first time to decide what constitutes “possession” of a drug within the meaning of our criminal statutes. Reeves, 209 N.W.2d at 21. We identified three elements the State is required to prove: (1) the defendant exercised “dominion and control (i.e., possession) over the contraband”; (2) the defendant had knowledge of the presence of the drug; and (3) the defendant knew the substance was an illegal drug. Id. at 21, 23.
After reviewing rules from other states, we concluded in Reeves that the State is not required to prove actual possession; proof of constructive possession will suffice. Id. at 21-24. In other words, the “dominion and control” of the drug required for possession does not mean the drug “needs to be found on [the defendant’s] person.” Id. at 22. It is enough that the defendant “maintains control or a right to control” the drug. Id. Control of the drug or the right to control it in the absence of actual possession constitutes constructive possession. Id.
The court also discussed in Reeves the circumstances under which such control or right to control, i.e., constructive possession, may be proved. We concluded that there are two factual scenarios under which knowledge of the presence of drugs and the ability to *636maintain control over them may be inferred. “If the premises on which such substances are found are in the exclusive possession of the accused,” the jury may infer constructive possession. Id. at 22, 23 (emphasis added). We specifically noted that joint possession of the premises was not sufficient to support an inference of control or knowledge of the drugs. Id. at 23; accord United States v. Morrison, 991 F.2d 112, 115 (4th Cir.1993); White v. State, 886 S.W.2d 876, 877 (Ark.Ct.App.1994); State v. Mercado, 887 S.W.2d 688, 691 (Mo.Ct.App.1994); Puente v. State, 888 S.W.2d 521, 526 (Tex.Ct.App.1994); see generally Short Annotation, 56 A.L.R.3d at 957-58 & Supp.1994 at 67-68 and cases cited therein.
The second situation in which we recognized an inference of possession was “when the contraband is found in a place which is immediately and exclusively accessible to the accused and subject to his [joint or sole] dominion and control.” Reeves, 209 N.W.2d at 22 (emphasis added); accord State v. Wiley, 366 N.W.2d 265, 270 (Minn.1985) (requiring exclusive access); Marshall v. State 110 Nev. 1328, 885 P.2d 603, 606 (1994) (applying same rule stated in Reeves)-, see Villegas, 871 S.W.2d at 896 (where accused is not in exclusive control of place where contraband found, there must be additional evidence to prove possession). It is important to note that this second scenario focuses on the place where the drugs are found as contrasted to the premises in which they are found.
These two rules and the differences between them are logical and reasonable. When we decided Reeves, we apparently believed that if drugs are “immediately and exclusively” accessible to the defendant, it is fair to impute knowledge and control of the drugs to the accused even though the accused shares dominion and control over the place or premises where the drugs are found. Stated conversely, we concluded that if the facts merely show joint dominion and control over the place or premises where the drugs are found, an imputation of knowledge and control would not be fair unless the defendant has immediate and exclusive access to the drugs.
These rules strike a balance between two competing considerations. We want to convict persons guilty of drug offenses even though they are not caught “red-handed” with the drugs on their person. State v. Florine, 303 Minn. 103, 226 N.W.2d 609, 610 (1975) (“purpose of constructive-possession doctrine is to include within the possession statute those cases where the state cannot prove actual or physical possession”); Charles H. Whitebread & Ronald Stevens, Constructive Possession in Narcotics Cases: To Have and Have Not, 58 Va.L.Rev. 751, 761-62 (1972). However, we also want to protect innocent bystanders from being convicted for drug offenses merely because they were in the wrong place at the wrong time. See Reid v. State, 212 Ga.App. 787, 442 S.E.2d 852, 853 (1994) (mere presence at the scene of a crime is insufficient to support a conviction of possession); State, v. Reynaga, 643 So.2d 431, 436 (La.Ct.App.1994) (same); Eaglin, 872 S.W.2d at 336 (same). I think the principles set forth in Reeves accomplish both objectives.
Unfortunately, we have not been entirely faithful to these principles. The instruction approved by the majority in this case illustrates how far we have strayed. I think we should either acknowledge that we have modified the rules enunciated in Reeves and articulate a reason for doing so or we should stand by those rules and reverse the defendant’s conviction. Because I am convinced that the law set forth in Reeves is sound, I would reverse the defendant’s conviction. The legal basis for my position follows.
In this case the court correctly instructed the jury that the State had to prove that the defendant knowingly possessed marijuana and that the defendant knew that the substance he possessed was marijuana. The court also correctly instructed the jury that possession could be actual or constructive and that constructive possession is shown by proof that the defendant knows of the presence of the drug and has the authority or right to control the drug. The jury was also correctly told that control of the drug could be joint or sole.
The problem in this case arose when the court added instruction no. 20 in an attempt *637to explain the circumstances under which a right to control the drug, i.e., constructive possession, could be inferred. The court instructed the jury that “[constructive possession occurs when the defendant maintains control or has a right to control the place where the controlled substance is found.” The court goes on to say that constructive possession may be inferred “when the substance is found in a place which is accessible to the defendant and is subject to his dominion and control, or the joint dominion and control of the defendant and another.” My interpretation of this instruction is that the jury can find the defendant controlled the drug or had the right to control the drug, i.e., constructively possessed the drug, upon proof that the drug was found in a place that was accessible to the defendant and proof that this place was subject to his joint or sole dominion and control.
I agree with the defendant that instruction no. 20 erroneously allowed the jury to find the defendant possessed the marijuana upon proof that the defendant jointly controlled the house in which the marijuana was found. At a minimum, the court should have given the defendant’s requested instruction that constructive possession could not be proved by showing that the drugs were found on premises over which the defendant had joint dominion and control. Without such an instruction the jury could easily conclude that the “place” where the drugs were found was the house, not some more localized site within the house/premises.
That the jury could so interpret the instruction on constructive possession is well illustrated by the comments of the trial judge and the county attorney which showed that they interpreted the instruction in this way. The additional instruction requested by defendant’s counsel stated that the jury could not infer knowledge and control of the contraband upon mere proof that the defendant had joint control of the premises. The county attorney resisted giving defendant’s requested instruction because it was “inconsistent with paragraph two of [the constructive possession instruction]” and the two instructions together would not make sense. The trial court agreed with these comments, noting that the requested instruction was inconsistent with the court’s instruction on constructive possession. The defendant’s attorney responded by stating that paragraph two of the court’s instruction was “an incorrect statement of the law.”
He was right and so were the county attorney and the trial judge. The defendant’s requested instruction was inconsistent with the court’s instruction. The court’s instruction allowed the jury to infer possession upon proof that defendant had joint dominion and control of the house where the marijuana was found. The defendant’s requested instruction told the jury just the opposite — possession could not be inferred from joint dominion and control; exclusive control of the house is required. If we are true to our holding in Reeves, we must agree with defendant that the court’s instruction on constructive possession was wrong. That is because in Reeves we held that “where the accused has not been in exclusive possession of the premises but only in joint possession, knowledge of the presence of the substances on the premises and the ability to maintain control over them by the accused will not be inferred but must be established by proof.” Reeves, 209 N.W.2d at 23 (emphasis added).
The harm flowing from the court’s incorrect instruction and refusal to give the defendant’s requested instruction is aggravated by the fact that the court’s instruction was incorrect for an additional reason. As noted earlier, the court instructed the jury that constructive possession could be inferred by proof that the drug was found “in a place which is accessible to the defendant” and subject to his sole or joint dominion and control. The original statement of this rule in Reeves required proof that the place was “immediately and exclusively accessible” to the defendant. Reeves, 209 N.W.2d at 22; accord State v. Pierce, 240 N.W.2d 678, 679 (Iowa 1976) (applying standard that constructive possession may be imputed when the drugs are found “in a place which is immediately and exclusively accessible to the accused”) (emphasis added). By eliminating the “immediate and exclusive access” requirement, it becomes even more crucial for the jury to understand that mere proof of *638joint control of the premises is not sufficient to infer knowledge and control of drugs found on the premises.
The instruction on constructive possession was wrong.2 The trial court was adequately alerted to this problem by defense counsel’s objections and requested instruction. Therefore, I would reverse and remand for a new trial. See Olson, 482 N.W.2d at 216 (erroneous instruction on constructive possession required reversal even though there was enough circumstantial evidence to support a conviction of possession of marijuana). I would instruct the trial court on remand to give only Uniform Jury Instruction 200.47, the possession instruction. The trial court should not instruct on inferences but instead should allow the attorneys to argue the evidence in support of their positions.
LAVORATO, NEUMAN, and SNELL, JJ., join this dissent.
. The majority correctly points out that "instructions on certain inferences may involve selective comments on the evidence by the trial court.” This criticism is particularly true in the context of constructive possession. Whether constructive possession exists depends on the peculiar facts of each case. State v. Harris, 647 So.2d 337, 339 (La.1994). Many factors may support a finding that an accused had knowledge of the presence of drugs and the right to exercise control over them in addition to access and control of the place and premises where the drugs are found. See Villegas v. State, 871 S.W.2d 894, 896-97 (Tex.Ct.App.1994). Other factors include incriminating statements made by the defendant, incriminating actions of the defendant upon the authorities’ discovery of the drugs, the location of the drugs among or near the defendant’s personal belongings, the defendant's fingerprints on the packages containing the contraband and any other circumstance which links the defendant to the drugs. See Emile F. Short, Annotation, Conviction of Possession of Illicit Drugs Found in Premises of Which Defendant Was in Nonexclusive Possession, 56 A.L.R.3d 948 (1974) (hereinafter "Short Annotation”). Any inference of possession should be drawn from all of the circumstances. Eaglin v. State, 872 S.W.2d 332, 337 (Tex.Ct.App.1994). Thus, the wiser approach is to avoid giving an instruction on the inferences to be drawn from selected evidence. State v. Olson, 482 N.W.2d 212, 216 (Minn.1992) (instruction on evidence giving rise to inference of constructive possession reversible error because it “was not a balanced instruction on the various relevant factors bearing on the jury's determination of the disputed possession issue but rather was one which singled out and unfairly emphasized one factor, one piece of the circumstantial evidence, bearing on that determination-”).
. To the extent that State v. Rudd, 454 N.W.2d 570 (Iowa 1990), is inconsistent with this conclusion, we should overrule it.