Determining an accurate method of obtaining complete morbidity data is a long-standing challenge for epidemiologists. The authors compared the accuracy and completeness of existing California hospital discharge data with self-reports of recent hospitalizations and surgeries from participants in the California Teachers Study. Self-reports were collected by questionnaire in 1997 from 91433 female teachers and administrators residing in California. Of the 13430 hospital discharge diagnoses identified for these women, cohort members reported 58%. Self-reporting was highest for neoplasms and musculoskeletal and connective tissue diseases and was most accurate for scheduled admissions, more recent admissions, longer lengths of stay, and less severe disorders. Hospitalizations for mental health and infectious disease were not as well reported. Among the 26383 self-reports-including outpatient surgeries, which are not captured by the hospital discharge database-confirmation was lower, as expected, especially for disorders of the nervous system and sense organs and skin and subcutaneous tissue. Confirmation was highest for childbirth admissions. The hospital discharge database was more specific, but the self-reports were more comprehensive, since many conditions are now treated in outpatient settings. The combination of self-reports and secondary medical records provides more accurate and complete morbidity data than does use of either source alone.