THE HYPOTHESIS THAT PUBLIC EDUCATIONAL EXPENDITURES ARE RELATED TO THE RELIGIOUS COMPOSITION OF A COMMUNITY WAS TESTED STATISTICALLY. EIGHTEEN RANDOMLY SELECTED SAN FRANCISCO BAY AREA SCHOOL DISTRICTS FROM A TOTAL OF 130 HAVING MORE THAN 300 PUPILS IN ADA WERE USED FOR THE STUDY. QUESTIONNAIRES ON RELIGIOUS PREFERENCES WERE SENT TO 50 RANDOMLY SELECTED VOTERS IN EACH DISTRICT. NINETY-ONE PERCENT USABLE RESPONSES WERE RECORDED. THE DATA WERE EVALUATED BY A MULTIPLE REGRESSION ANALYSIS AND TESTED FOR SIGNIFICANCE BY USING A "T" TEST. A MULTIPLE CORRELATION COEFFICIENT PROVIDED A MEASURE OF THE USEFULNESS OF THE REGRESSION EQUATION FOR PREDICTING THE CRITERION VARIABLE. INDEPENDENT VARIABLES FOUND TO BE MOST RELATED TO EDUCATIONAL EXPENDITURES WERE THE PERCENTAGE (IN THE COMMUNITY) OF (1) CATHOLIC, (2) PROTESTANT-NURTURE, (3) PROTESTANT-CONVERSION, AND (4) MISCELLANEOUS. THIS STUDY MAKES FOUR CONTRIBUTIONS TO THE UNDERSTANDING OF DETERMINANTS OF SCHOOL EXPENDITURES--(1) UTILIZATION OF A VARIABLE NOT PREVIOUSLY EXAMINED, (2) EMPLOYMENT OF A METHOD OF DATA COLLECTION USEFUL FOR FUTURE STUDIES, (3) CATEGORIZATION OF RELIGIOUS DENOMINATIONS INTO VARIABLES WHICH CAN BE EXAMINED STATISTICALLY, AND (4) INITIAL STATISTICAL EXAMINATION OF THESE VARIABLES. THIS PAPER WAS PREPARED FOR THE AMERICAN EDUCATIONAL RESEARCH ASSOCIATION (CHICAGO, FEBRUARY 11, 1965). (HW)