OBJECTIVE: This study tested whether a relationship exists between concentration and response following discontinuation of selective serotonin reuptake inhibitors. METHOD: Eight patients with remitted major depression who were taking 20 mg/day of either fluoxetine or paroxetine underwent placebo substitution for 3 days. Serum drug and brain fluorine levels were obtained before and after placebo substitution. RESULTS: With placebo substitution, a mean of 88% (SD=13%) of brain fluorine signal from fluoxetine (plus fluorinated metabolites) remained, compared with a mean of 38% (SD=17%) of the brain fluorine signal from paroxetine (plus fluorinated metabolites). Among patients taking paroxetine, adverse events during placebo substitution correlated highly with steady-state brain drug levels. CONCLUSIONS: The correlation of clinical effects with brain drug levels in the paroxetine group suggests that relationships between drug response and brain drug concentrations merit further investigation.