Although the principal brain target that all antipsychotic drugs attach to is the dopamine D2 receptor, traditional or typical antipsychotics, by attaching to it, induce extrapyramidal signs and symptoms (EPS). They also, by binding to the D2 receptor, elevate serum prolactin. Atypical antipsychotics given in dosages within the clinically effective range do not bring about these adverse clinical effects. To understand how these drugs work, it is important to examine the atypical antipsychotics' mechanism of action and how it differs from that of the more typical drugs. This review analyzes the affinities, the occupancies, and the dissociation time-course of various antipsychotics at dopamine D2 receptors and at serotonin (5-HT) receptors, both in the test tube and in live patients. Of the 31 antipsychotics examined, the older traditional antipsychotics such as trifluperazine, pimozide, chlorpromazine, fluphenazine, haloperidol, and flupenthixol bind more tightly than dopamine itself to the dopamine D2 receptor, with dissociation constants that are lower than that for dopamine. The newer, atypical antipsychotics such as quetiapine, remoxipride, clozapine, olanzapine, sertindole, ziprasidone, and amisulpride all bind more loosely than dopamine to the dopamine D2 receptor and have dissociation constants higher than that for dopamine. These tight and loose binding data agree with the rates of antipsychotic dissociation from the human-cloned D2 receptor. For instance, radioactive haloperidol, chlorpromazine, and raclopride all dissociate very slowly over a 30-minute time span, while radioactive quetiapine, clozapine, remoxipride, and amisulpride dissociate rapidly, in less than 60 seconds. These data also match clinical brain-imaging findings that show haloperidol remaining constantly bound to D2 in humans undergoing 2 positron emission tomography (PET) scans 24 hours apart. Conversely, the occupation of D2 by clozapine or quetiapine has mostly disappeared after 24 hours. Atypicals clinically help patients by transiently occupying D2 receptors and then rapidly dissociating to allow normal dopamine neurotransmission. This keeps prolactin levels normal, spares cognition, and obviates EPS. One theory of atypicality is that the newer drugs block 5-HT2A receptors at the same time as they block dopamine receptors and that, somehow, this serotonin-dopamine balance confers atypicality. This, however, is not borne out by the results. While 5-HT2A receptors are readily blocked at low dosages of most atypical antipsychotic drugs (with the important exceptions of remoxipride and amisulpride, neither of which is available for use in Canada) the dosages at which this happens are below those needed to alleviate psychosis. In fact, the antipsychotic threshold occupancy of D2 for antipsychotic action remains at about 65% for both typical and atypical antipsychotic drugs, regardless of whether 5-HT2A receptors are blocked or not. At the same time, the antipsychotic threshold occupancy of D2 for eliciting EPS remains at about 80% for both typical and atypical antipsychotics, regardless of the occupancy of 5-HT2A receptors. The "fast-off-D2" theory, on the other hand, predicts which antipsychotic compounds will or will not produce EPS and hyperprolactinemia and which compounds present a relatively low risk for tardive dyskinesia. This theory also explains why L-dopa psychosis responds to low atypical antipsychotic dosages, and it suggests various individualized treatment strategies.