Exp. I involved an analysis of ratings given to members of contrasting word association pairs. Exp. II and III involved an analysis of many-into-one paired-associate learning employing the words hot, warm, mild, cool, and cold as response terms. Exp. I indicated that members of oppositional pairs differ markedly in their locations on a series of 3 rating scales, while Exp. II and III revealed that pairs involving hot and cold as response terms were learned more readily, produced fewer errors, and were given less frequently as incorrect guesses than any of the other 3 words in the series. Results are discussed in terms of their implications for contiguity and generative theories of word association production. (15 ref.) (PsycINFO Database Record (c) 2006 APA, all rights reserved)