Abstract Background: Deficits in set shifting have been identified in schizophrenia, reflecting rigidity in problem solving, for example, not using feedback to change behavior. The aim of this study was to examine potential long-term development of set-shifting abilities for schizophrenia patients and their associations with psychopathology, both at baseline (BL) and at follow-up (FU). Methods: In a longitudinal follow-up study, we reexamined 33 initially antipsychotic-naive patients with schizophrenia and 43 healthy controls 7–16 years after illness onset, on their set-shifting abilities using the Intra-Extra Dimensional Set Shifting Task from Cambridge Neuropsychological Test Automated Battery. Positive, negative, and general symptoms in the patient group were assessed with the Positive and Negative Syndrome Rating Scale (PANSS) at BL and FU. We used a univariate analysis of covariance in a preliminary analysis of change scores on both extra-dimensional shift (EDS) errors and total errors (adjusted for stages not completed) with the following covariates: age, sex, baseline level performance, and the time interval between BL and FU examinations (7–16 years). Difference between PANSS subscores (positive, negative, and general) at BL and FU scores was assessed with repeated measures test. We used Spearman’s correlation analysis between PANSS composite scores and EDS errors, and total errors adjusted at BL and FU. Results: We found a significant group difference in change scores of EDS errors (P = .025) and total errors adjusted (P = .049). Patients showed increased performances (less errors) when compared to controls who remained relatively stable (mean EDS error change for patients = −3.2 (SD = 10.8) and controls = −0.7 (SD = 4.2) and mean total errors adjusted change for patients = −5.1 (SD = 23.3), controls = −1.5 (SD = 8.3)).There was a significant decline in PANSS positive scores (BL = 19.7 (SD = 4.3), FU = 14.4 (SD = 5.5), t(31) = 4.9, P < .001), negative scores (BL = 21.6 (SD = 4.6), FU= 17.5 (SD = 6.2), t(31) = 3.3, P = .002), and total scores (BL = 77.6 (SD = 12.2), FU = 65.2 (SD = 1.6), t(31) = 3.6; P = .001) and a nonsignificant decrease in PANSS general scores (BL = 36.3 (SD = 7.9), FU = 33.4 (SD = 7.9), t(31) = 1.48 (P = 0.15)). We did not find significant correlation between symptom dimension and EDS errors or total errors adjusted, neither at BL nor at FU. Conclusion: The preliminary results show a longitudinal increase in set-shifting abilities in the years following onset of schizophrenia compared to a relative stability of set-shifting abilities in the healthy control group. Set-shifting deficits could not be linked to psychopathology. Support of work: The study was supported a by grant (R25-A2701) from the Lundbeck Foundation. The funding bodies had no role in the design or conduct of the study.