This paper examines the robustness of Deaton’s widely used method for estimating consumer responses. While unit values, ratios of expenditures to quantities purchased, are often employed in demand analysis as proxies for missing market prices, Deaton argues bias is likely to result as a consequence of both quality effects and measurement error. Hence he proposes a procedure that corrects the bias and enables price elasticities to be obtained in the absence of explicit price information. Given the availability of market price data and unit values in Vietnam, this paper estimates a food demand system and investigates the usefulness of Deaton’s method. It also takes the analysis a step further to the existing literature by computing the welfare impact of price changes to see how Deaton’s method performs in this context. The results demonstrate that Deaton’s method generates materially different price elasticities from those estimated with market prices. However, it produces relatively similar results for the welfare analysis. Deaton’s procedure therefore appears to be valid in welfare analysis at least in the case of Vietnam, but the findings also indicate that its use by policymakers should carry a strong health warning.