THE READING ACHIEVEMENT OF FOURTH-GRADE STUDENTS WHO WERE TAUGHT WORD RECOGNITION BY EITHER A BASAL APPROACH OR A BASAL APPROACH WITH LINGUISTIC EMPHASIS WAS INVESTIGATED. A SAMPLE OF 10 CLASSROOMS MATCHED ON INTELLIGENCE, CHRONOLOGICAL AGE, AND SOCIOECONOMIC STATUS WAS SELECTED FROM SCHOOL SYSTEMS IN IOWA AND MICHIGAN. THE LORGE-THORNDIKE NONVERBAL INTELLIGENCE TEST, THE BOND, CLYMER, HOYT SILENT READING DIAGNOSTIC TEST, AND THE GATES READING SURVEY WERE ADMINISTERED. ANALYSIS OF COVARIANCE, T TESTS, AND MULTIPLE REGRESSION WERE USED TO ANALYZE THE DATA. UNADJUSTED MEAN SCORES INDICATED THAT THE EXPERIMENTAL GROUP WAS SUPERIOR IN ALL SKILLS EXCEPT WORD SYNTHESIS. WHEN INTELLIGENCE WAS USED AS THE COVARIATE, THE EXPERIMENTAL GROUP WAS SUPERIOR. THE AVERAGE AND LOW MENTAL ABILITY GROUPS BENEFITED MORE FROM THE EXPERIMENTAL TREATMENT THAN DID THE HIGH ABILITY GROUP. THE FOLLOWING THREE WORD RECOGNITION SKILLS CONTRIBUTED SIGNIFICANTLY TO RAPID COMPREHENSION--SYNTHESIS, RECOGNITION OF WORDS IN CONTEXT, AND PHONETIC KNOWLEDGE. MOST OF THE MULTIPLE CORRELATIONS INDICATED THAT WORD RECOGNITION SKILLS ACCOUNTED FOR AT LEAST ONE HALF THE VARIANCE IN PREDICTING READING ABILITY. FURTHER RESULTS, TABLES, AND A BIBLIOGRAPHY ARE INCLUDED. THIS ARTICLE APPEARED IN THE "READING RESEARCH QUARTERLY," VOLUME 1, NUMBER 4, SUMMER 1966. (BK)