Normal subjects were given 54 Kent-Rosanoff word-association- test items in one of two different orders; antonym-eliciting items were concentrated either (a) near the beginning or (b) near the end of the list. For each order, testing was administered under three different test conditions: (a) standard free-association instructions, (b) instructions to give the response "most people" would give, and (c) "most people" instructions with a multiple- choice test format. The order starting with antonym-eliciting items elicited more popular antonym responses than did the other order. Popularity-set instructions, particularly with the multiple-choice format, elicited more non-antonym popular responses than did free-association test conditions. With repeated testing, popular antonyms became more frequent. For some sequences of test conditions, there was also an increase in non-antonym popular responses with repeated testing.