Using detailed cost breakdowns, this study attempts to determine the extent to which achievement of fourth- and eighth-grade students can be predicted from educational costs and school district size, and what combination of these variables and socioeconomic status best predicts student achievement (as measured by 1980-81 California Achievement Test scores in 152 school districts in a Southeastern state). Reported here with 15 tables are, first, a comparison of two canonical correlation models indicating the importance of socioeconomic status relative to cost and district size variables; and second, multiple regression analysis of the correlations of 10 predictor variables with test scores in each of 7 achievement areas at both grade levels. The findings indicate that costs of instruction are significantly related to student achievement. The authors conclude, however, that the cost of supporting a given level of achievement in low-income populations is quite high, and that mathematics concepts and applications scores are higher in large school districts due to greater teacher specialization in those districts. (MCG)