Nine sets of annual data on State school finances are used to test a theory of expenditure determination by public school districts. The results support implications of the theory regarding effects of personal income, State and federal aid, the relative price of education, the pupil/population ratio, and enrollment growth on per pupil spending. A population density variable and a South versus non-South regional variable, both included on the basis of earlier results, also affect spending significantly. The nine cross-sectional equations are generally consistent, but there are some structural shifts over time, and the hypothesis of coefficient homogeneity is not supported. Consequently, a pooled equation that allows for such shifts provides the most useful predictive model. Policy applications of the results are limited by (1) omission of some "taste variables" that affect spending, (2) uncertainty about differential State responses to aid, and (3) the absence of price data for individual States. (Author)