This article uses administrative data from Manitoba and New England to address the reasons underlying Manitoba's relatively high mortality in the 30 days after hip fracture repair. Both the Manitoba and New England data sets are population based, containing information on individuals 65 years of age or older in Manitoba (1979-1992; n = 10,007) and New England (1984-1985); n = 16,206). Various logistic regression models were estimated on pooled and separate data from Manitoba and New England; the models all showed similar predictive accuracy, having C statistics in the .71 to .74 range. Manitoba postsurgical 30-day mortality rates were greater than the 1984 to 1985 New England rate for each of the 14 years considered. In particular, New England residents with very short waits before hip fracture repair (0 or 1 day) had mortality rates both markedly lower than expected and significantly less than those of Manitobians with such short waits. Attention to the Manitoba hospitals with very poor 30-day survival and to the process surrounding selection of patients for early versus late surgery in Manitoba are clearly in order. The extent to which longer-term survival reflects 30-day survival also is discussed. Our findings highlight the utility of comparative data for understanding quality of care problems within a single region.