The clock-drawing test (CDT) is used widely to evaluate cognitive disorders, but its role in the assessment of psychotic disorders has not been studied. We sought to examine whether the CDT plays a role as an indicator of psychosis and to establish its sensitivity to clinical improvement of psychosis. The CDT was administered twice to 53 hospitalized patients without dementia but with psychosis: once at admission and again before discharge. The CDT scores were calculated in a random order by two independent senior psychiatrists who were blinded to the patients' status (admission or discharge). The inter-rater reliability was high (0.89 at admission, 0.85 at discharge, P<0.01 for both). The severity of psychosis was assessed by the Positive and Negative Syndrome Scale (PANSS). Patients had significantly lower CDT scores at admission than at discharge (2.87±1.39 vs. 3.91±1.08, respectively, P<0.01). The PANSS-total score of the patients showed a significant improvement (84.90±17.77 vs. 69.18±16.23, P<0.01). An inverse correlation was found between CDT performance and psychosis severity, as reflected by the PANSS-positive symptom subscale at admission (R=-0.279, P<0.05). Our findings suggest that the CDT may aid in the assessment of psychotic states and in their clinical monitoring.