Six American universities were studied to assess the predictive validity of Scholastic Aptitude Test (SAT) scores and other admissions data and of student language background relative to college grades for Hispanic students. Sample sizes at the universities ranged from 76 to 686 for Hispanic subjects and from 924 to 4,919 for White students. The universities included three institutions (two in California and one in Texas) with Hispanic students of predominantly Mexican American extraction, one institution in Florida with a majority of Cuban American students among the Hispanic sample, and two institutions (one in Massachusetts and one in New York) with a majority of Puerto Rican students among their Hispanic samples. Data sources included: (1) College Board's Admissions Testing Program/Validity Study Service files; student records on freshman grades and, sometimes, high school grades, standardized high school rank, and/or college major; and (3) student responses to two special surveys on language background for Hispanic subjects. Results of multiple regression analysis show no evidence of lower accuracy of prediction for Hispanic students either for the tests alone or in combination with other variables. In examining the relationship between language background and predictor variables, it appears that SAT verbal scores correlated moderately with English proficiency, but SAT math scores were less correlated with English proficiency. The results support the use of SAT scores in combination with the high school record for evaluation of Hispanic American students in admissions to moderately and highly selective institutions. The accuracy of prediction varies by institution, but differences between ethnic groups at the same institution are small. Additional methodological notes and questionnaires and communications with the students are appended. (TJH)