PURPOSE: The aim of the study was to investigate whether the most recent introduced atypical antipsychotics olanzapine and risperidone were preferentially prescribed to patients susceptible to develop extrapyramidal side effects (EPS) and those notresponding adequately to typical antipsychotics. METHODS: Data were obtained from the Dutch PHARMO system that includes complete medication and hospital admission records of 675 000 residents of 14 Dutch cities. A total number of 129 new users of olanzapine and 142 new users of risperidone as well as 507 new users of typical antipsychotic drugs were identified from our database in the period of 1996-1998. The prevalence of markers of EPS, therapy resistance and therapy non-compliance were assessed in the period of 1 year prior to a new start of an antipsychotic. RESULTS: New use of olanzapine and risperidone was significantly associated with previous use of other antipsychotics (odds ratio 4.0, 95%CI: 2.5-6.7 and odds ratio 3.0, 95%CI: 2.0-4.7, respectively). New use of olanzapine and risperidone was also associated with previous use of anticholinergic drugs compared to users of typical antipsychotics (over three and two times more, respectively). This effect diminished when adjusted for previous useof antipsychotics. CONCLUSIONS: In particular olanzapine and also risperidone were selectively prescribed to patients formerly treated with other antipsychotics and to those susceptible for EPS. If not recognised or controlled for, observational studiescomparing different antipsychotic drugs may produce biased results on efficacy or frequency of side effects for the different types of antipsychotics.