The validity of studies on the diagnostic significance of first-rank symptoms (FRS) for schizophrenia has been put in doubt because of a poor compliance with Schneider's criterion for their definition and the lack of use of the phenomenological method for their assessment. In this study, using a rigorously phenomenological approach to elicit FRS, we examined (a) the degree to which unequivocally present FRS differentiated schizophrenia (n=513) from other psychotic disorders (n=633), and (b) the comparative validity between FRS and other reality-distortion symptoms against 16 external validators in the whole sample of psychotic disorders (n=1146). Diagnostic performance indices (with 95% CIs) of FRS for diagnosing schizophrenia were as follows: sensitivity=0.58 (0.54−0.61), specificity=0.65 (0.62−0.67), positive predictive value=0.57 (0.54−0.60) and negative predictive value=0.65 (0.63−0.68). While the overall association pattern of FRS and non-FRS scores with the validators was rather similar, three validators (premorbid social adjustment, number of hospitalizations and global assessment of functioning) were significantly related to non-FRS scores (p < 0.006) but not to FRS scores (p > 0.05). Furthermore, no validator was significantly related to FRS scores and unrelated to non-FRS scores, all of which indicates an overall better predictive validity for non-FRS delusions and hallucinations. These findings suggest that FRS do not have diagnostic value for diagnosing schizophrenia and that they do not meaningfully add to the external validity showed by other delusions and hallucinations. We believe that much of the misunderstanding about the diagnostic and clinical validity of FRS for schizophrenia is rooted in Schneider's confusing concept of the disorder.