Subjects were given problems each consisting of a stimulus word with its five, seven, or nine strongest word associates. They attempted to guess the middle-strength (target) associate. Since the hierarchy of strength between a stimulus and its associate responses is such that differences in strength among associates become progressively smaller as strength between stimulus and response decreases, associates weaker than the target associate are more similar in strength to the target than are associates stronger than the target. It was therefore expected that in trying to guess the target associate, more errors would be made to weaker than to stronger associates. This was found in all conditions. More generally, it appeared that the farther removed in strength--whether stronger or weaker-an associate was from the target associate, the less it was interfering. Giving the subjects a cue to associates stronger or weaker than the target word had some effect on, but in no way eliminated, the strong tendency to make more errors on weaker associates. In three previous studies it has been found that when subjects attempted to solve problems that required guessing the nth item in a verbal-response hierarchy of five members, items of intermediate habit strength were more difficult to guess than either stronger or weaker members of the hierarchy. The problems used have been (a) stimulus words each with its five strongest word associates,' (b) sets of five bigrams where the bigrams within a set were of different frequencies of occurrence in words,2 and (c) sets of five words, the words in a set having different frequencies of occurrence in the printed language.3 In all of these studies, interest was cenReceived for publication June 11, 1970. The study was supported in part by Grant HD-00901 from the Institute of Child Health and Human Development, National Institutes of Health, United States Public Health Service. Thanks are due Donald Lehr and Robert Koch. SC. P. Duncan, Response hierarchies in problem solving, in C. P. Duncan (ed.), Thinking: Current Experimental Studies, 1967, 18-41. 2R. L. Dominowski, Problem difficulty as a function of relative frequency of correct responses, Psychon. Sci., 3, 1965, 417-418. N. H. Jurca and C. P. Duncan, Problem solving within a word-frequency hierarchy, J. verb. Learn. verb. Behav., 8, 1969, 229-233.