There are now more than 50 studies concerning neuroleptic blood levels and clinical outcome relationships. Haloperidol, the most studied, is the only antipsychotic permitting some conclusions. A number of authors suggest that the striking lack of agreement between different studies results from heterogeneity of their quality. Here, we have proposed a score system for assessing the quality of those studies. According to it, none (0/14) of the studies having a score <0.60 were able to show a therapeutic window, as compared to 53% (10/19) of those having a score >0.60 (P=0.002, Fisher exact test). Also, the studies able to identify the presence of a therapeutic window during haloperidol treatment were those having sample sizes >20 (P=0.60) and those whose patients were treated with fixed doses (P=0.02). The diagnosis of schizophrenia in the studies does not seem to be an exclusive condition, as compared with those also including schizophreniform and schizoaffective disorders (P=0.12). In the stepwise logistic regression, only fixed dose and sample size of the studies were found to significantly explain the presence or absence of a therapeutic window. Our qualitative analysis of haloperidol blood level publications seem to indicate that an upper limit may exist for haloperidol efficacy; values above this limit seem not to provide any supplementary improvement and may even reduce therapeutic effect.