In order to determine whether the kind of process underlying cloze responses is indeed a systematic and exhaustive search, a study was conducted exploring some corollaries to such a search hypothesis. It was assumed that subjects would generate responses representing a number of word types, that some of these word types would be sensible and some nonsensical, and that responses would be representative of the entire body of possible response words. Five versions of a 300-word cloze passage, every fifth word deleted, were administered to 390 junior-high-school students who were randomly assigned to one of the versions. Protocols were hand scored and success probabilities were calculated. A correlation matrix among seven variables was calculated and analyzed using a stepwise regression program. Significant correlations were noted among the seven variables, with the highest correlation appearing between size of response body related to success probability. It was concluded that the general search hypothesis appeared to be sustained since distribution of responses was related to success probability and since the ratio of nonsense to sensible responses was relevant to that distribution. It was implied that a search process could be characterized as systematic in part. Tables and references are included. (MS)