This study attempts: (1) to verify that serum antimuscarinic activity is related to clozapine dose, and more importantly to clozapine plasma concentrations; (2) to explore whether norclozapine has serum antimuscarinic activity; (3) to explore whether antimuscarinic activity is related to clozapine side effects; and (4) to compare the serum antimuscarinic activities of clozapine with those of antiparkinsonian drugs and other antipsychotics. In 39 patients participating in a double-blind clozapine study, the [3H]QNB assay was used to measure serum antimuscarinic activity: (1) on baseline medications; (2) after a 4-week haloperidol trial; (3) after a 16-week clozapine trial of either 100, 300, or 600 mg/d; and (4) after 1 or 2 consecutive 16-week clozapine trials with remaining doses in nonresponders. Clozapine levels predicted serum antimuscarinic activity better than clozapine dose. At the end of the 1st clozapine trial, the correlation with the levels explained 69% of the variance of serum antimuscarinic activity (r = 0.83, P < 0.001, N = 34). Clozapine levels were good predictors of serum antimuscarinic activity only in patients taking 300 or 600 mg/d. After correcting for clozapine levels, the within-subject correlation between norclozapine levels and serum antimuscarinic activity was relatively high and significant (r = 0.54, F = 26.7, df = 1.65, P < 0.001). Constipation was significantly associated with higher serum antimuscarinic activity during the 1st clozapine trial. Clozapine was associated with clearly higher antimuscarinic activity than other antipsychotics or low doses of antiparkinsonians. In vitro studies and new clinical studies are needed to verify whether norclozapine may significantly contribute to antimuscarinic activity during clozapine treatment.