The investigation was concerned with the class of grapheme-phoneme correspondence called spelling patterns and was intended to broaden empirical understanding of grapheme-phoneme correspondences used in teaching initial reading. Twenty-five boys and 47 girls from three moderate-ability first grade classes were chosen as subjects. These students had accumulated six minutes of session time on the computer assisted reading program (CAI) phonic strand before beginning the experiment. One set of 24 spelling patterns was devised for each of the five vowels: A, E, I, O, and U. When each subject began the experiment, he was randomly assigned to one of the five sets of spelling patterns and then 12 of the 24 spelling patterns in the chosen set were selected for training. At the end of his eight-day training period, each subject was individually tested, off-line, on all 144 initial consonant and final consonant combinations derived from the full set of 24 spelling patterns to which he was assigned. The data reflected the value of presenting spelling patterns in initial reading. Teaching spelling patterns to the subjects resulted in positive transfer to a criterion task that required subjects to read words and non-words composed of the spelling patterns taught. (WR)