Although therapeutic drug monitoring of clozapine is recommended, its optimisation is often adjusted only on the basis of dosage. The aim of this study was to assess the link between clozapine plasma concentrations and clinical response by a meta-analysis of published studies and by an individual participant data meta-analysis. We conducted a computerised search of bibliographic databases (EMBASE, PubMed, Clinical Trials, and Web of Science) to identify studies that assessed the relationship between clozapine serum or plasma concentrations and clinical efficacy. Using pooled data, we investigated the association between improvement of clinical outcome and clozapine or norclozapine plasma concentrations, the sum of clozapine and norclozapine plasma concentrations, and the coefficient of variation of clozapine plasma concentrations. Using available individual data, we assessed the relationship between clozapine plasma concentrations and clinical response (changes in the Brief Psychiatric Rating Scale score) and identified a threshold level for a favourable clinical response. Fifteen studies satisfied inclusion criteria. Our meta-analysis showed that responders had clozapine plasma concentrations that were, on average, 117 ng/mL higher than non-responders. The patients with plasma clozapine concentrations above the thresholds identified in each study had a higher likelihood of responding (odds ratio = 2.94, p < 0.001). Norclozapine plasma concentrations were not associated with a clinical response. The meta-analysis of individual data supported this result and confirmed the link between clozapine concentrations and a change in the Brief Psychiatric Rating Scale score and/or the probability of clinical response. Finally, with the analysis of the coefficient of variation of clozapine plasma concentrations, we found that a greater inter-individual fluctuation in plasma concentrations was associated with a loss of clinical response. Our work confirmed that, in contrast to clozapine doses, clozapine plasma concentrations were related to a favourable clinical response, with a mean difference between responders and non-responders of 117 ng/mL. A threshold for a treatment response of 407 ng/mL was determined, with a high discriminatory capacity,and a sensitivity and specificity of 71% and 89.1%, respectively.