Choice between alternative flexible functional forms has received little explicit treatment in many empirical agricultural studies. Theoretical criteria and empirical techniques for choosing between flexible functional forms are reviewed. Theoretical topics include definitions of flexibility, mathematical expansions, separability, and regular regions. Empirical techniques examined are Monte Carlo analysis, parametric modeling, bayesian inference, and nonnested hypothesis testing. Comparison of the full range of theoretical and empirical aspects may provide more credible and reliable empirical estimates when consumer or producer duality assumptions are appropriate in agricultural applications.