OBJECTIVES: Studies of small area variations of health care utilization are more and more frequent. Such variations are often considered to be an indication of variations in the quality of medical care. The variations in the rate of operations for hip fractures are among the lowest studied to date, due to the fact that a consensus exists concerning this surgery. Our objective is to examine these variations within the context of relatively small and heterogeneous districts. METHOD: Based on anonymous computerized data on public hospital stays, this study describes the variations in population rates (crude and standardized) of operations for hip fracture among the health districts of the Canton of Vaud for the period from 1986 to 1991. District populations vary from 22,000 to 164,000. Using the extremal quotient (EQ), the importance of these variations was determined. RESULTS: The study population consists of 2363 cases, of which 78% are women. Mean age is 80.4 for women and 70.6 for men. Standardized rates of operation for hip fracture per 100,000 in the Canton Vaud for the years 1986 to 1991 are, respectively: 56; 67; 86; 91; 89 and 94. The EQ for the years 1986 to 1991 are respectively: 8.2; 4.0; 3.5; 2.7; 1.9 and 1.9. The high EQ, especially for the earlier years, are contrary to the initial premise of absence of variation. The progressive implementation in the Canton Vaud of VESKA medical statistics could play a role, as could the small size of many of the districts, with resultant instability of rates. CONCLUSIONS: Considering the wide variations shown here for an operation hardly regarded as subject to variations, it is important to exercise caution in interpreting published data of small area variations.