We examined short- and long-term changes in neurocognitive functions in patients with schizophrenia who were either started or switched to amisulpride in comparison with the normal controls. Fifty-seven patients treated with amisulpride and 60 normal controls completed a comprehensive neurocognitive function test battery at the baseline, the 8-week, and the 1-year follow-up. We conducted and compared the results of both intention-to-treat (ITT) and per-protocol (PP) analyses to account for the follow-up loss. Three general results obtained were as follows: (1) the degree of the improvements in neurocognitive function was comparable to those of other second-generation antipsychotics in both ITT and PP analysis; (2) in light of the relative effect size, the composite effect size and the effect size in most measures in both ITT and PP analyses were smaller for the patient group than those of the control group, signifying that improvement in performance may be largely attributable to practice effects; and (3) nonetheless, there were evidences of both short- and long-term improvements in some cognitive tasks, such as in the Korean-Wechsler Adult Intelligence Scale vocabulary subtest and the Trail Making Test, that may not be accounted by practice effect. These results suggest the need to include a healthy control group to validate the medication effect of cognitive improvements in patients with schizophrenia and to consider practice effect in interpreting the results of repeated administration of neurocognitive function tests.