Backward masking deficits have been put forward as potential psychological markers for vulnerability to schizophrenia. This study was conducted to investigate whether schizophrenic patients improve their performance on a backward masking task during a single test session. The ability of a degraded stimulus version of the masking task to act as a specific diagnostic marker for paranoid schizophrenia (versus affective disorder) was also investigated. The backward masking task was performed on 18 paranoid schizophrenic patients, 18 unipolar depressed patients, and 18 non-psychiatric controls. Paranoid schizophrenic patients were included because they tend to show normal performance with traditional masking protocols. Schizophrenic patients made significantly more detection errors compared to depressives and non-psychiatric controls where interstimulus intervals (ISIs) longer than 14 ms were used. Unlike depressed patients and non-psychiatric controls, schizophrenic patients showed no reduction in error rate during the entire period over which the backward masking task was performed. The constant error rate which was observed at an ISI of 114 ms suggests that schizophrenic patients cannot attenuate the disruption effect due to deflection of attention from the target to the mask. The backward masking deficit in schizophrenia appears to arise from a temporarily stable visual processing impairment in performance within a single test session.