Spanish and English versions of a 38-item grammar test were administered to 40 Spanish-surnamed pupils equally divided by sex at grade levels K, 1, 3, and 5 (10 subjects per grade) in a bilingual education program. The test was a revision of part of an earlier test for oral proficiency in Spanish and English. The reliability of the new test was .95 for the Spanish version and .96 for the English version; but many items failed to elicit the expected grammatical constructions. English scores increased significantly with grade level, while Spanish scores varied only slightly among grade levels. Balance between mean scores on the English and Spanish versions was reached at grade 3. Test subjects were also questioned about language use and preference. The relationships between reported language use and test results appeared to indicate that Spanish proficiency was determined by use of the language in the home. English proficiency showed some relationship to use with the peer group. There were no significant correlations between the English and Spanish version scores and only very slight relationships among the factors influencing them. Achievement in English thus appeared to be unrelated to the maintenance of Spanish for bilingual children. (Author)