The development accounting literature almost always assumes a Cobb-Douglas (CD) production function. However, if in reality the elasticity of substitution between capital and labor deviates substantially from 1, the assumption is invalid, potentially casting doubt on the commonly held view that factors of production are relatively unimportant in accounting for differences in labor productivity. We use international data on relative factor shares and capital-output ratios to formulate a number of tests for the validity of the CD assumption. We find that the CD specification performs reasonably well for the purposes of cross-country productivity accounting.