It has been established that memorizing common problems and their solutions underlies cognitive skill development, and that there are substantial age deficits in the rate of this learning. In a between-groups design, the authors compared learning rates for the same set of problems in skill (SK) training and paired-associate (PA) training. The authors found main effects due to condition (PA problems were acquired earlier) and to age (older adults' learning was delayed), but no condition-by-age interaction. The authors concluded that the age deficit in SK can be accounted for by the age deficit in associative memory; no further explanation is needed. The authors also analyzed fast and slow retrieves in SK and PA, and found that the frequency of fast retrieves did not differ in the two conditions. The overall advantage of PA was due to the occurrence of slow retrieves, which were absent in SK presumably because the skill algorithm displaces slow, explicit memory search in SK, but not fast, familiarity-based retrieval.