Measures of capital services are used in studies of production and to inform policies related to growth and development. A variety of methods have been used to measure capital stocks and service flows. We briefly review the methods commonly used to measure capital service flows, and the main assumptions. We then quantify the substantial differences between our newly constructed InSTePP series on capital use in U.S. agriculture and a comparable USDA series. We show that measures of capital services are sensitive to the treatment of interest rates, notably the use of fixed versus variable market rates, and we demonstrate the implications for measures of the quantity and productivity of agricultural capital in the United States. We conclude that when calculating capital usage in U.S. agriculture the use of a fixed rate of interest will generate more plausible estimates than the use of an annual market rate that varies from year to year. Capital Services in U.S. Agriculture: Concepts, Comparisons, and the Treatment of Interest Rates "The capital time series is one that will really drive a purist mad." Robert Solow (1957, p. 314) An accurate measure of the annual flow of capital inputs is valuable for policy makers and researchers who are interested in production and productivity. However, estimates of capital stocks and service flows are difficult to calculate and vulnerable to significant measurement errors because of data limitations and the myriad of assumptions required. Estimates of the flow of capital services are especially sensitive to underlying assumptions. Information about the implications of the alternatives provides a basis for making better-informed choices about the appropriate approaches and assumptions to apply when measuring capital stocks and flows.This article begins with a review of methods used commonly to measure capital stocks and service flows, making explicit a number of important assumptions required to construct such measures. Next, we examine and compare the measures of capital inputs in U.S. agriculture from two contemporary, state-specific panel data sets. We compare the methods used to construct the capital series, and we reveal and discuss differences in data sources, the types of data used to construct the capital measures, and the resulting estimates. We also outline some assumptions about depreciation, service lives, interest rates, aggregation methods, and the scope of goods included in each of the capital series for each data set. Finally, we examine and illustrate the extent to which certain choices