This study addressed the importance of medical record information in determining the presence of psychotic symptoms in first-admission patients. A sample of 232 first-admission inpatients screened for psychotic symptoms by facility personnel was administered the Structured Clinical Interview for DSM-3-R (SCID) followed by a medical record review and interview with a significant other. Medical records were unavailable for sixteen patients. These patients were more often female, married, and more suspicious than patients whose records were reviewed. Of those having interviews and record reviews, 49 showed no clear evidence of psychosis. The remainder were divided into three groups: 97 subjects who revealed all of their psychotic symptoms during the interview (SCID-ALL); 61 who revealed only some of their delusions or hallucinations during the interview (SCID-PART); and 25 who revealed none of this information during the interview but whose records clearly described psychosis (SCID-NONE). The three groups were reasonably similar demographically and with respect to clinical history. Clinically, at the time of interview, SCID-NONE subjects were less often still psychotic, were rated on the Brief Psychiatric Rating Scale as less depressed, more withdrawn, less cooperative and less severely ill, and had poorer insight ratings on the Hamilton Depression Scale.