Clinical studies of monoamine neurotransmitter function in depression have concentrated on individual monoamines without focusing on interactions between monoamine systems. Virtually all modern studies have found significant correlations between monoamine metabolite concentrations in cerebrospinal fluid (CSF). These correlations should in part reflect interactions between central monoamine systems. In the present analysis, CSF had been obtained from depressed patients before (n = 40) and after (n = 36) antidepressant treatment. The patients were grouped based on their response to treatment. Absolute concentrations of CSF monoamine metabolites (homovanillic acid, 5-hydroxyindoleacetic acid, and 3-methoxy-4-hydroxyphenylethyleneglycol) did not differ between the two groups before or after treatment. However, when correlations between metabolites were compared, nonresponders to treatment differed considerably from responders. In responders, as in previously described normal populations, all three metabolites correlated with one another before and after treatment, and treatment-induced changes in metabolite concentrations also correlated with one another. In contrast, metabolites in nonresponders did not correlate with one another before treatment, nor did treatment-induced changes correlate with one another in this group. Furthermore, correlations between treatment-induced changes in metabolites differed significantly between responders and nonresponders, and there was a trend for pretreatment correlations to differ as well. The lack of correlation between monoamine metabolites in nonresponders suggests that interactions between monoamine systems may be disrupted in these individuals. Using CSF metabolite correlations to study neurotransmitter interactions may have clinical relevance and yields information not available from examining neurotransmitters in isolation.