Animal models are crucial for understanding the mechanism of action of antipsychotics. However, the dose of an antipsychotic in animal studies is often arbitrarily chosen, with haloperidol 1 mg/kg being a rather common standard. Recent clinical positron emission tomography (PET) studies in patients show all antipsychotics to block dopamine D2 receptors, and most are effective at doses that lead to 60% to 80% D2 occupancy. When occupancy exceeds 80%, the incidence of side effects rises sharply. To use this "bedside" information to inform the "bench," we measured D2 occupancy in rats using a method similar in principle to the [11C]-raclopride PET method in humans. We found that: 1) as in humans, haloperidol is effective in animal models of antipsychotic action when D2 occupancy > 70% and leads to effects in models of extrapyramidal side effects when D2 occupancy is > 80%; 2) very low doses, 0.06 mg/kg/sc, cause acute D2 occupancy of 75%; 3) and even doses that acutely saturate D2 receptors give little D2 occupancy after 24 hours due to the very short half-life of haloperidol in rats (2.5 hours versus 24 hours in humans). We conclude that most previous animal studies of antipsychotics have used doses giving rise to inappropriately high acute D2 occupancy and inappropriately low D2 occupancy between doses. We exemplify how this dosing confounder could lead to inappropriate conclusions. Data from the bedside translated to the bench--using D2 occupancy as a mediating variable--will lead to more valid animal models.