Summary Six paired-associate lists, four constructed on the basis of associative strength and two on the basis of semantic distance of pairs of response members, were studied with regard to amount of intralist interference. A control list with response members remotely related with respect to these two variables was used for comparison. The types of relationships between six experimental response words were: (1) antonyms, (2) coordinate words, (3) strong direct associates in free association which are neither antonyms nor coordinates, (4) words which are not strong associates but which have high proportions of common responses in a free-association test, (5) connotatively, as well as denotatively, similar words, and (6) only connotatively similar words. The results showed that among the four association lists, only the lists of strong associates and antonyms were more difficult than the control list. Of the two semanticdistnace lists, only the synonym list was more difficult than the control list. Learning the control list first facilitated the learning of an experimental list significantly more than vice versa. The value of associative strength and semantic distance as predictors of interferences in paired-associate learning was discussed.