Abstract Objective To assess the effects of second generation antipsychotics on neurocognitive function in patients with stable remission of first episode psychosis. Methods Fifty-three patients with first onset psychosis in the schizophrenia spectrum entered a randomised controlled trial of guided discontinuation (GD) versus maintenance treatment (MT) with second generation antipsychotics. A comprehensive neurocognitive test battery was administered at the time of remission and shortly after dose reduction or discontinuation (GD-group) or at the same time in the MT-group. Results With the exception of negative symptoms, PANSS scores decreased over time and neurocognition improved significantly on most tests in both groups. The GD-group, however, improved significantly more than the MT-group on three neurocognitive measures in the domain of speed of processing. Conclusion These data suggest that, in first episode patients, dose reduction or discontinuation of second generation antipsychotics after stable remission is achieved, might improve neurocognitive function more than continuing second generation antipsychotics, suggesting a negative role for second generation antipsychotics, specifically in the domain of speed of processing.