Numerous double-blind studies have compared second-generation antipsychotics (SGAs) with first-generation antipsychotics (FGAs), with most finding better efficacy and tolerability for SGAs. However, these 'efficacy trials' were generally short term and included only highly selected patients. Mostly because of weight gain and other metabolic effects of the SGAs, as well as their high acquisition price, the debate on the (cost) effectiveness of the SGAs led to two pragmatic clinical trials with no sponsorship by industry. Both trials had broad inclusion criteria and long follow-up, and tried to mimic clinical routine: CATIE (Clinical Antipsychotic Trials of Intervention Effectiveness) and CUtLASS (Cost Utility of the Latest Antipsychotic drugs in Schizophrenia Study). 1493 patients participated in CATIE, an 18-month, double-blind trial comparing the SGAs olanzapine, quetiapine, risperidone and ziprasidone with the FGA perphenazine. If efficacy or tolerability was insufficient, patients were re-randomized to a medication other than the one they previously received. Improvement of psychopathology and of quality of life was only moderate. Overall, 74% of patients discontinued study medication before 18 months, and the median time to discontinuation was 4.6 months. Aside from olanzapine (time to discontinuation 9.2 months), the other SGAs did not differ from each other or from perphenazine. Except for adverse effects as a reason for discontinuation, differences between the SGAs and the FGA were minimal. In CUtLASS, a 12-month open-label trial, 277 patients were randomized to receive an FGA or a SGA. Again, efficacy was rather similar between the two groups, with only limited improvement of psychopathology and quality of life. The authors of both trials concluded that SGAs do not markedly differ from FGAs regarding compliance, quality of life and effectiveness. The methodological problems of both trials have been discussed extensively. Patients had psychotic symptoms that were moderate in severity and were at least partially treatment resistant. The marginal improvement observed indicated that this population might not be appropriate to detect differences between FGAs and SGAs. Specific issues of CATIE include the exclusion of patients with tardive dyskinesia in the perphenazine arm and the high discontinuation rate. In CUtLASS, the concept of including 13 different FGAs and four SGAs in the respective classes was problematic. It is of interest that the most widely prescribed drug was sulpiride--of the FGAs, this is probably the 'most atypical' drug. Aside from the finding that the advantages of the SGAs are not as strong as early trials and marketing suggested or promised, the trials do not provide much helpful information regarding everyday practice. For tardive dyskinesia, no conclusions at all can be drawn. Similarly, methodological problems inhibited the detection of the other major advantage of the SGAs, i.e. the improved subjective well-being/quality of life while receiving these agents. It is well known that patients' and doctors' perspectives differ markedly, and the Quality of Life Scale (QLS), an expert-rated scale used in both trials, might not be sensitive enough to detect the subjective advantages reported by the majority of patients in other trials. CATIE and CUtLASS suggest that SGAs do not live up to all the previous expectations. However, even if most of these advantages are debatable, the lower risk of tardive dyskinesia and the better subjective effects should be strong enough reasons to favour these drugs. There is no single antipsychotic that is best for every schizophrenia patient, as individual responses differ markedly. For successfully individualized treatment, a multitude of antipsychotic options are needed.