The aim was to determine the post-onset longitudinal course of cognitive functioning in first-episode psychoses and to examine how premorbid adjustment, duration of untreated psychosis (DUP), and clinical variables such as relapse are associated with that course. Consecutive patients with a DSM-IV diagnosis of non-organic psychosis coming to their first treatment in the health care areas under study were included. Ultimately, 207 patients were assessed neuropsychologically at baseline, 138 were reassessed one year later, and 111 two years later. Five dimensions were identified through principal component analysis of eight neuropsychological (NP) test results: Working Memory (WM), Executive Function (EF), Verbal Learning (VL), Impulsivity (Im), and Motor Speed (MS). No major changes were found in the level of neurocognitive functioning from baseline to the 1-year and 2-year follow-ups. Patients with good initial levels of premorbid academic functioning had consistently better scores on WM at all three time points. No association was found between DUP and the longitudinal course of neurocognitive function. Significant associations occurred between better WM and VL at 1 and 2 years and fewer relapses during the first year, but not the second. Most NP deficits are in place by onset of psychosis and are stable over two years. Milder WM deficits are associated with higher premorbid academic functioning. More severe deficits in WM and VL are associated with more relapses during the first year. It is unclear whether NP deficits cause relapse, relapse causes NP deficits, or both are manifestations of a third deteriorative process.