The frequency distribution of platelet 5-HT(2A) receptor densities (measured as [3H]LSD B(max)) was analysed in 60 patients with major depression and 40 matched control subjects to determine whether the observed receptor densities come from a single distribution or whether there are more distributions that are represented differently in control subjects and in diagnostic subgroups of non-suicidal and suicidal patients. The distribution of B(max) values in all subjects was tested by using the NOCOM program. The analysis has shown a best fit for a trimodal distribution of values (low, intermediate and high binding). There were significant differences in average probabilities of control subjects and patients from the two diagnostic subgroups belonging to any of the three distributions. In the control and non-suicidal groups, a significantly higher (P=0.003) proportion of individuals had a probability of belonging to a low binding distribution. In contrast, the probability of belonging to the high distribution was significantly greater (P=0.007) in the suicidal group of patients than in the non-suicidal group or in control subjects. In all three groups, the proportion of cases in each distribution fit those expected under the Hardy-Weinberg equilibrium. The results support the notion that high 5-HT(2A) receptor density is a marker of suicidality, possibly genetically determined.