OBJECTIVES: Understanding factors associated with the average cost of providing critical care, particularly size of unit, is important for many policy decisions. Study of possible economies of scale, i.e. decrease in average cost with increased size of unit, has been limited due to a lack of valid cost data collected on a uniform basis. This study uses a cost database in which data from a group of 75 U.K. critical care units are held, collected using consistent definitions. It estimates the extent of scale economies controlling for other relevant critical care unit characteristics. METHODS: Annual costs for 75 British critical care units were obtained using a cost accounting system based on specific costs of resource inputs and prices. Cost categories included staff costs, consumables and clinical support services. Multiple regressions using cost per patient day as the dependent variable and log of unit size (number of beds) as the major independent variable were performed. Separate regressions were estimated for each of the cost categories as well as for total costs. RESULTS: Median unit size was 6 beds (IQR 4.9 to 8 beds). Median total cost per patient day was £955 pounds (IQR £884 to £1105 pounds) A statistically significant negative relationship between total cost per patient day and unit size was found (p<.001, adjusted R-squared=.33). The size coefficient was also negative in the regression for each cost component, and was statistically significant in the equations for staff and consumables costs. CONCLUSIONS: Average costs per patient day in British critical care units decline with increased size of unit. The liklihood of scale economies should be considered in facilities planning, cost projection and budgeting, and economic evaluation of policy proposals.