This study examined the assumption that language expression and reading performance are related processes. Subjects included a total of 676 nine-year-old children of heterogeneous socioeconomic status, intelligence, and achievement levels. Verbal fluency was defined as being a measure of associative verbal encoding (a/v/e), wherein children give single word associations to a stimulus word. Two hypotheses were examined: (1) a/v/e would improve with training, and (2) associated with a/v/e would be concomitant improvement in reading achievement. Training in a/v/e involved encouraging subjects to give as many responses as possible to the stimulus words. Two training periods were held daily and conducted by the regular classroom teacher. Only two verbal stimulus words were presented at any one time. In the control classes an equivalent time block was devoted to general language arts instruction. Subjects were unaware that an experiment was in progress. Results showed the experimental group improved in a/v/e, vocabulary, comprehension, and total reading performance. It was also determined that a/v/e could be reliably quantified and measured. (HOD)