A common observation is that measures of productivity growth are pro-cyclical, meaning they are higher (or grow faster) on average during periods of economic expansion than during periods of economic contraction. This study focuses on measurement errors related to capital inputs as an explanation for the existence of pro-cyclical patterns in measures of agricultural productivity. Calculating a time series of capital inputs is difficult and prone to errors. Myriad assumptions are required to construct a typical measure of the capital stock, and further, sometimes related, assumptions must be made about the utilization of the stock to derive a measure of capital service flows. We test the hypothesis that unmeasured changes in the utilization of capital can affect productivity measures. This is accomplished using recently constructed indexes of inputs, outputs, and productivity in U.S. agriculture for 1949-2002. We find that utilization responses by farmers are significant and bias measures of productivity growth in a pro-cyclical pattern. The bias is quantified and the measures of productivity are adjusted for the estimated utilization responses and compared to the original measures.