This paper deploys the Granger method to determine the directions and patterns of causality between national income and total as well as various components of public expenditure. Empirical results based on the data for India (1950-81) suggest that while at the disaggregate level the causal process is rather diverse, it is essentially feed back-type at the aggregate level. It neither confirms the Wagnerian (Income Public Expenditure) nor the Keynesian (Public Expenditure National Income) view. The paper recommends that two variables be treated as jointly-dependent in both the public finance and macroeconometric studies.