BACKGROUND: The ability to measure and track changes in risk-adjusted obstetric complication rates using administrative data underpins efforts to improve obstetric quality of care, but the validity of this approach has not been adequately evaluated. We sought to assess the validity of using composites of ICD-9-CM codes to identify selected categories obstetric complications and risk factors associated with complications. METHODS: Patients with ICD-9-CM codes for obstetric trauma/laceration, infection, haemorrhage, episiotomy or obesity discharged between January 2009 and March 2010 were identified in the study hospital's administrative data. One hundred medical records with ICD-9-CM codes of interest were randomly selected for review from each of the five categories. An additional 60 medical records without the ICD-9-CM codes of interest served as controls for each category. Sensitivity and specificity for the selected categories was estimated using inverse proportional weighting to adjust for sampling based on presence of one of the ICD-9-CM codes of interest. RESULTS: Weighted sensitivities ranged from 0.15 [95% CI 0.11, 0.20] for obesity to 1.00 for overall infection while specificities ranged from 0.994 [95% CI 0.987, 0.998] for obesity to 0.999 [95% CI 0.996, 1.000] for episiotomy. Obese patients were not reliably identified and it was not possible ascertain whether some diagnoses were present on admission. CONCLUSIONS: For selected categories of obstetric complication diagnoses, use of composite sets of ICD-9-CM codes may be a valid method to identify patients within these complication categories. © 2012 Blackwell Publishing Ltd.