Lexical decision tasks were used in 17 schizophrenic subjects and 11 control subjects, to evaluate the so-called automatic priming of semantic links, and the controlled processes which involve longer lexical anticipation strategies. In this type of trial, inhibition in the recognition of the target word, when the preceding priming word is not semantically linked to it, indicates the activation of such controlled processes. Postulating that patients in the schizophrenic group were heterogeneous in terms of cognition, we compared the resultant scores in two subtypes of schizophrenic patients: paranoid and hebephrenic (CIM-9 criteria). The study demonstrated that there was a difference in performance between the schizophrenic subtypes. Although these findings need to be confirmed, they demonstrate the necessity of determining individual profiles of patients entering the trial, to take into account those of their phenomenological differences that may correspond to differences in cognitive functioning. The results indicate an absence of inhibition, but only in the paranoid group. In these subjects, results were consistent with the hypothesis of a failure to set up controlled processes rather than a stronger “automatic” activation of the semantic links. However, as the results were not significant, the hypothesis remains open to further investigation. The reasons for the nonsignificance of the results and alternative hypotheses are discussed.