This paper examines the implications of extending the Ahmad-Stern model of indirect tax reform to include labour supply. The inclusion of labour supply alters the basic measure of marginal revenue cost (MRC) of indirect taxation and introduces the possibility of calculating a MRC for direct taxation. The paper derives the expressions for these revised MRCs and provides estimates from Irish data. It then examines the sensitivity of the results to assumptions regarding functional form and, in particular, goods/leisure sc'parability. In setting optimal tax rates policy-makers face considerable difficulties. Explicit assumptions must be made regarding the underlying preferences of the consumer (i.e. the functional form for the utility and demand functions) and also about how expenditures are distributed. Sensitivity to underlying preferences is discussed by Deaton (i 98 I) and is empirically examined by Ray (I986) and Ebrahimi and Heady (I988). Policy-makers also face the problem that even when they do provide explicit forms for demand functions they face difficulties regarding the reliability of estimates and are also evaluating demand responses at a point possibly quite far away from the optimum. The difficulties outlined above led to the development of the marginal tax reform literature, with the seminal paper in this area by Ahmad and Stern (I984). This approach has the considerable advantage of not requiring the choice of explicit utility functions, nor of distributions of expenditure, but instead merely requiring information on the actual position of the economy at a single point in time, using actual consumption and distributions of expenditure. It must be noted that tax reform does require information on demand responses, so that results in this area can still be sensitive to the specification of consumer preferences and demand systems. However, tax reform results appear to be less sensitive than do tax design results to this