We have reported an association between schizophrenia and homozygosity of a Bal I polymorphism in the first exon of the dopamine D3 receptor gene (Crocq et al.: Journal of Medical Genetics 29:858-860, 1992). The present study consists of an attempt to replicate this finding in a further sample of 66 patients and 97 controls. Once again more patients than controls were homozygous, but the effect was not as strong as in our first study (chi 2 = 2.53, P = 0.05, one tailed). When pooled data from our two studies were analysed, excess homozygosity in patients remained highly significant (P = 0.002) with a particular excess of the 1:1 genotype (P = 0.01). This reflected a departure from Hardy-Weinberg equilibrium in the patients (P = 0.0005) but not the controls (P = 0.24). This led us to explore the possibility that there might be important differences between the patients in our two studies and that excess homozygosity might be a characteristic of particular subgroups of schizophrenics. Our findings suggest that the effect is consistently at its strongest in those patients who have a high familial loading and in those who have a good response to neuroleptic treatment, and that differences between our two samples might have contributed to the quantitatively different outcomes.