Depressed and nondepressed Ss were exposed to either inescapable noise or no noise conditions in an attempt to replicate a typical learned helplessness study (Miller & Seligman, 1975) with Israeli student nurses. The inescapable noise was presented either as a personally important task or as an unimportant task. When later tested on a series of 20 patterned anagrams, no performance differences were found between depressed and nondepressed Ss as was found in previous studies, nor did the importance manipulation affect performance. Within the depressed group only, it was found that the more the Ss believed they had control over the noise, the better was their performance in the anagrams. Similarly, only among the depressed was the rated aversiveness of the noise positively related to anagram performance. Depressed were only able to assess accurately the number of failures on the anagrams, while nondepressed did equally well in assessing both positive and negative aspects of their performance. The results are discussed in terms of Seligman's learned helplessness model of depression and Beck's cognitive model of depression.