Previous studies had revealed no specific effect under haloperidol (typical) and risperidone (atypical) neuroleptic (NLP) treatments for schizophrenia (SZ) on a variety of neurocognitive functions relying on the dopaminergic meso-cortico-limbic system (Rémillard et al., 2005, 2008). Considering the different affinities of D2 dopamine receptors for typical and atypical NLPs, these drugs may differentially affect the functions of the striatum, a determinant brain structure involved in procedural learning. The influence of risperidone (2–6 mg) and haloperidol (2–40 mg) on a nonmotor procedural task involving semantically related pairs of words with inverted letters was investigated in this double-blind study. The performance of 26 patients with SZ, randomly assigned to risperidone or haloperidol, was compared to that of 18 healthy controls at baseline, 3, 6, and 12 months. Results revealed that all patients with SZ exhibited slower reading speed of the word pairs than healthy controls at all assessment periods. In addition, procedural learning – characterized as a significant decrease in the time taken to read aloud the target word pairs – was more impaired in the haloperidol- than in the risperidone-treated group at all assessment periods. Healthy controls showed steady improvement in reading speed over the 12 months of the study, in contrast to SZ patients, who reached a plateau in their capacity to improve mirror-reading skill over time. However, all SZ participants in the study showed near normal learning profiles from exposure to semantic associations embedded in the procedural memory task, providing evidence for the preservation of associative connections in the semantic network of these patients. The observed impairment in procedural learning in SZ may thus reflect, at least in part, the influence of neuroleptic medication on striatal functions. (JINS, 2010, 16, 148–156.)