The objective of the current study was to compare the performance of schizophrenic patients and normal controls on implicit memory tests. Two neuropsychological tasks were administered to 29 patients and normal participant samples. The implicit tests were: Word fragment completion and Word production from semantic categories. The priming score was the variable of interest. Priming effects are obtained in normal subjects and schizophrenia patients, regardless of the implicit test used. However, a dissociation in priming between normal and patient groups was observed, depending on the test used. For word fragment test, priming was identical between the two groups. However, for word production, priming obtained in schizophrenics was lower than priming in normal controls. Results confirm a dissociation effect in implicit memory tests. These results could be explained in the context of the Roediger and Blaxton (1987) distinction between data-driven and conceptually-driven processing. This evidence suggests that a complete neuropsychological assessment of memory in schizophrenia should include different kinds of implicit memory tests (procedural, perceptual, and conceptual tasks).