The literature on art productions from psychiatric patients consistently suggests that specific graphic characteristics distinguish among various major diagnostic groups. These hypothesized differences are summarized. Systematic collection of art productions as part of separate research programs studying affective psychoses and schizophrenia has enabled us to amass a collection of psychotic patients' drawings which are used in an attempt to evaluate these hypotheses empirically. Art sessions were held with 104 patients hospitalized for affective psychoses and with 62 patients hospitalized for acute schizophrenia at NIH's Clinical Center. Comparing these drawings across diagnostic groups revealed, to our surprise and contrary to extant hypotheses, considerable within-diagnostic group variability and between-group overlap. There were, nonetheless, some trends in the hypothesized directions. These differences, however, disappeared when comparisons were made on a subsample of age-matched patients. This discrepancy between this research and expectations based on the literature is discussed. Our studies indicate that pictorial characteristics are not closely associated with diagnosis. We have, however, found that individual patient's art productions, and their associations to the pictures, are of great value in arriving at a dynamic understanding of the patient, regardless of diagnosis.