This paper provides a methodological bridge leading from the well-developed theory of credit rationing to the less developed territory of empirically identifying credit constraints. We begin by developing a simple model showing that credit constraints may take three forms: quantity rationing, transaction cost rationing, and risk rationing. Each form of non-price rationing adversely affects household resource allocation and thus should be accounted for in empirical analyses of credit market performance. We then outline a survey strategy to directly classify households as credit unconstrained or constrained and, if constrained, to further identify which of the three non-price rationing mechanisms is at play. We discuss several practical issues that arise due to the use of a combination of “factual” and “interpretative” survey questions. Finally, using a data set from northern Peru, we demonstrate the importance of accounting for all three forms of credit constraints by estimating the increase in farm production that would result from relaxing credit constraints. The inclusion of transaction- and risk-rationed households in the constrained group results in an estimated impact that is twice as large as the impact when only quantity rationed households are considered constrained.