To compare radioimmunoassay of postpartum maternal hair samples with a structured maternal interview for the detection of cocaine use during pregnancy. Radioimmunoassay of hair samples obtained postpartum was compared with self-report of cocaine use obtained by confidential, structured interviews of 405 postpartum women. Cocaine or benzoylecgonine was detected in the hair samples of 129 of the 148 women (87%) who reported using cocaine at least once during pregnancy. Thirty-six of 257 women (14%) who reported that they had used no cocaine during pregnancy had positive hair tests. Positive hair assay with negative self-report was significantly more common among women who were unmarried (P = .001), African-American (P < .001), and multiparous (P = .035). Hair analysis is a valuable complement to maternal self-report by interview. The methods used together provide a useful clinical and research technique yielding a comprehensive estimate of gestational cocaine exposure. Either method alone may result in misclassification of gestational cocaine exposure.