The study prepared reading tests in the children's first language, Spanish, that initially considered their comprehension of the language and subsequently evaluated their skill in drawing inferences or pertinent information from the material read. Two reading comprehension tests (Forms A and B) were constructed, using two different excerpts from "Fantasias", a book not previously read by the children. Each test consisted of 35 multiple-choice questions divided into 3 parts: Input Subtest--20 questions assessing the student's familiarity with a passage's linguistic elements; Content Subtest--5 items measuring the student's ability to infer meaning from context; and Paraphrase Subtest--assessing the ease with which students recognize suitable paraphrases for 10 of the verbal expressions found in the Input passage. A total of 150 examinations were administered to 116 third-grade students (predominantly of Mexican and Puerto Rican backgrounds) enrolled in 6 bilingual schools in Milwaukee, Wisconsin. Five classes took Form A (79 tests), 4 took Form B (71 tests), and 2 took both forms (34 tests). Test instructions were given in both English and Spanish. Student scores were then compared with factors such as language dominance, home language, and reading levels through a series of correlations. Among the findings were that: reliability of Form A (total test) was high at .93; for Form B reliability was higher for subtests II and III and noticeably lower for subtest I; significant at the .01 level were the total score and subscores for Form A with the children's reading level in Spanish. (NQ)