Three reading procedures for learning isolated words--whole-word, syllable-blending, and phonics-blending--were examined to determine which was the most effective for preschool children. The six preschool children in the study were divided into two groups to assess the effects of the length and number of different letters in a series of training words on acquisition, maintenance, and generalization. The words were counterbalanced across procedures, subjects, and presentation positions. The results showed that the whole-word procedure resulted in the most rapid acquisition for recall, the fewest training errors, the fewest letter-order recognition errors, and the most frequent recall of training words in compound words, sentences, and stories. After the syllable- or phonics-blending procedures, the subjects performed better for both reproduction of individual letter-sounds and generalizations of letter-sounds to pseudowords. Regardless of the training procedure, all words were maintained equally. (FL)