This paper investigates the nature, extent, and impacts of credit constraints in Ethiopia’s agriculture. Using a direct elicitation approach on a panel of 5,308 smallholder farmers, we find that around 66.6% were credit constrained, a majority of them (71.9%) due to risk factors and transaction costs (14.33%). The hypothesized heterogeneity of credit constrained farmers is corroborated by the results from the endogenous regime switching regression model which show that the determinants of credit constraints and their impact on farm productivity are specific to the type of constraints farmers face. Predictions from this model indicate that alleviating credit constraints would generate substantial productivity gains in Ethiopia of around 60%. Our findings suggest that expanding farmers’ access to financial information, increasing the number of branch offices of banks and microfinance institutions in the country and particularly in rural areas, and easing financial transaction costs might increase farmers’ access to credit and significantly alleviate their credit constraints.