This exploratory study aims to examine the differential effects of a computer-based cognitive training in 'prodromal' patients (mean age 27.20 years, S.D. 5.31 years) compared with patients with full-blown schizophrenia (mean age 30.13 years, S.D. 7.77 years). Ten patients at risk for schizophrenia and 16 patients suffering from schizophrenia underwent a computerized cognitive training program (Cogpack). Cognitive functioning before and after a total of 10 training sessions was assessed by different tests controlling for memory, attention, and logical thinking. Prodromal patients turned out to be able to significantly improve their long-term memory functions and their attention after cognitive training with the Cogpack software package whereas in the group of patients with schizophrenia no improvement occurred (e.g. continuous performance test, identical pairs-subtest 'shapes': improvement from 0.73 to 0.88 in persons at risk of schizophrenia vs. no improvement in patients with schizophrenia (0.55 to 0.53). Cognitive training using Cogpack is helpful for the improvement of cognitive functioning in persons at risk of schizophrenia. Thus, the application of cognitive training should be provided as early as possible in the prodromal phases of schizophrenia in order to use the full rehabilitative potential of the patients. These results should be confirmed by further investigations including larger sample sizes.