Two experiments aimed at extending the principles of paired-associates learning transfer to the acquisition of reading skills are reported. Approximately 15 first graders were randomly assigned to each of the 10 treatment conditions. In the first experiment, four types of letter training were compared to two control conditions. The experimental conditions were (1) high phonic correspondence between letter name and sound (ABAcBc), (2) high mediating correspondence between letter name and sound (ABAcB'c), (3) low correspondence between letter name and sound (ABAcCc), and (4) first task observation of letter stimuli (AOAcBc). The control conditions were (1) warm up (ABCcDc) and (2) no letter training (--AcBc). Results showed a significant effect attributable to presentation method, with the ABAcBc and ABAcB'c tasks producing the greatest transfer and AOAcBc producing the least. The second experiment used preliminary training in mediational cues in ABAcBc to determine whether this would increase transfer to the second task. Three groups, ABAcBc, ABAcCc, and --AcBc, were compared. Results showed that training improved performance significantly. Conclusions reached after the two experiments were that (1) the ABAcBc paradigm produced maximum transfer, indicating that a phonics approach to reading is the most efficient and (2) mediation skills training is valuable in tasks such as these. References are included. (MS)