Objective There is no report that statistically evaluates the therapeutic reference (350–600 ng/ml) and adverse drug reaction (ADR) range (>1000 ng/ml) of clozapine (CLZ) recommended by the Arbeitsgemeinschaft für Neuropsychopharmakologie und Pharmakopsychiatrie (AGNP) consensus guidelines in an isolated and large sampling study. Methods We administered CLZ to 131 Japanese patients with treatment‐resistant schizophrenia in a multicenter cross‐sectional study. Plasma CLZ concentrations were assayed by high‐performance liquid chromatography using trough sampling. The Brief Psychiatric Rating Scale (BPRS) and severe dose‐dependent ADR (sedation, myoclonus, and seizures) were analyzed statistically after adjusting for possible confounders. Results The daily CLZ dosage showed a moderately positive relationship with the plasma concentration (r = 0.49, p < 0.001). Every 100 ng/ml increase in plasma CLZ concentration improved the total BPRS score 1.95% (95% CI: 0.89–3.01, p < 0.001) and the odds ratio (OR) 1.38 (95% CI: 1.14–1.66, p = 0.001) for BPRS response. Compared with concentrations below 350 ng/ml CLZ, 350–600 ng/ml (11.12%; 95% CI: 2.52–19.72, p = 0.012) and 600–1000 ng/ml (11.05%; 95% CI: 2.40–19.71, p = 0.013) showed significant improvement in the total BPRS score. Dosages above 1000 ng/ml showed greater improvement (25.36%; 95% CI: 13.08–37.64, p < 0.001) of the total BPRS score but more severe ADRs than dosages below 1000 ng/ml (OR: 31.72; 95% CI: 1.04–968.81, p = 0.048). Conclusion The AGNP therapeutic reference range (350–600 ng/ml) is useful, and a dose above 1000 ng/ml is potentially more effective but carries the risk of severe ADRs in the central nervous system.