Examines two independent data sets to support argument that although onset-rime sensitivity typically predicts school entrants' later word reading skills, phoneme sensitivity predicts more variation. Maintains that multiple regression analyses do not reveal level of phonological sensitivity needed to understand alphabetic reading instruction and describes a different analytic strategy for addressing the issue. (Author/KB)