<i>Background:</i> The diagnostic significance of first-rank symptoms (FRSs) remains uncertain. Ethnic differences in FRSs may account for high rates of schizophrenia in minority groups. This study aims to examine the prevalence of FRSs in an epidemiological sample of first-episode psychoses stratified by relevant demographic variables. <i>Sampling and Method:</i> We identified everyone aged 16–64 presenting with their first psychosis over 2 years in 3 UK centres. <i>Results:</i> A total of 426 subjects had consensus diagnoses of DSM-IV and ICD-10 psychotic conditions. Thirty-eight percent (95% CI = 33–42) reported FRSs; more frequent in those classified as having schizophrenia (DSM-IV: 55%, 95% CI = 47–63; ICD-10: 51%, 95% CI = 44–58) than those with affective psychoses (DSM-IV: 31%, 95% CI = 22–39; ICD-10: 29%, 95% CI = 21–38). FRSs in schizophrenia were more common in white British subjects, while in affective psychoses, they were more frequent in the black group. The sensitivities, specificities and positive predictive values for schizophrenia of FRSs were 55, 69 and 72% according to DSM-IV and 51, 71, 74% according to ICD-10, respectively. The sensitivities were higher in white British than in the black group. <i>Conclusions:</i> FRSs were common but unhelpful for differentiating schizophrenia from other psychoses as they occurred frequently in both diagnoses. Phenomenological differences did not explain the higher incidence of schizophrenia in black ethnic minority groups.