To compare methods of measuring negative symptoms, eight rating scales were employed to retrospectively assess and subtype 187 patients with schizophrenia from the Chestnut Lodge Follow-up Study. These included Andreasen's Schedule for Assessment of Negative Symptoms, Carpenter's Criteria for the Deficit Syndrome, Kay and Opler's Positive and Negative Symptom Scale, the scales developed by Krawiecka et al and Crow's modification of them, the Negative Symptom Scale developed by Lewine et al, Pogue-Geile and Harrow's Negative Symptom Scale, and Abrams and Taylor's Emotional Blunting Scale. The overlap and concordance, temporal stability, and predictive validity of these instruments are described. When rated from detailed medical records, the reliability of all scales was fair to good. Despite their inclusion of different items, there were high positive correlations between the scales when used to rate negative symptoms dimensionally. When used to classify individual patients as having the negative or deficit syndrome, however, concordance among criteria was low. Using the broadest criteria (Pogue-Geile and Harrow), 75 (40%) patients were diagnosed as having negative syndrome; the narrowest criteria (Andreasen and Olsen) yielded 11 (6%) diagnoses of negative syndrome. Narrower definitions tended to be subsets of broader ones. Carpenter's Criteria for the Deficit Syndrome focus on primary enduring negative symptoms and show the greatest temporal stability. Broader criteria, which diagnose the deficit or negative syndrome independent of severity of positive symptoms, had the greatest predictive validity.