Examination was made of a proposal by Underwood, Ekstrand, and Keppel (1965) concerning the effects of the various subprocesses that are assumed jointly to determine paired-associate learning. Runquist's (1968a, 1968b, 1969) recently presented distinction between formal and rated similarity was evaluated. Only minimal support was found for the Underwood, et al. predictions which involve the paired-associate learning subprocesses of response-learning and associative interference where formal similarity is concerned. Further, the use of rated similarity as a ratio-scale measure on intralist similarity appears to furnish a definition of formal similarity which is more rigorous than the manner by which it has been traditionally ordinally scaled.