The objective of this study was to assess the dose-response relationship of standard and higher doses of olanzapine in a randomized, double-blind, 8-week, fixed-dose study comparing olanzapine 10 (n = 199), 20 (n = 200), and 40 mg/d (n = 200) for patients with schizophrenia or schizoaffective disorder and suboptimal response to current treatment. Patients meeting criteria for antipsychotic treatment resistance were excluded. Dose-response relationship was assessed by linear regression analysis with log-transformed dose (independent variable) and Positive and Negative Syndrome Scale (PANSS) total score (dependent variable). There were no significant dose group differences in patients completing the study (overall, 67.8%). All dose groups showed statistically significant improvement in PANSS total scores from baseline to end point without significant dose-response relationship (P = 0.295). Post hoc analysis of response showed significant interaction between baseline PANSS and dose (P = 0.023), indicating better response at higher doses for patients with higher baseline PANSS. There was a significant dose response for mean change in weight (P = 0.003) with significant difference between the 10- and 40-mg-dose groups (P = 0.002; 1.9 [10 mg/d], 2.3 [20 mg/d], and 3.0 kg [40 mg/d]). There was a significant dose response for change in prolactin (P < 0.001) with a significant difference between each group (-10.5 [10 mg/d], -1.7 [20 mg/d], and 4.9 ng/mL [40 mg/d]; P < or = 0.018). Over 8 weeks, non-treatment-resistant patients with schizophrenia or schizoaffective disorder responded to all 3 doses of olanzapine, without a statistically significant dose-response relationship, suggesting that for many patients with schizophrenia or schizoaffective disorder, particularly those who are mildly or moderately ill, 10 mg/d should be the initial dose of choice.