Although response sets have been extensively investigated as determiners of individual responses to personality tests, their impact on educational tests has not been well studied. In this study, I investigated positional response bias, testwiseness, and guessing strategy as components of variance in test responses on multiple-choice tests of learning. University students (A r = 62) enrolled in a learning skills course responded to two content exams, a testwiseness measure, and a guessing strategy measure. Of the students, 6 showed a statistically significant tendency to mark early options. The proportion of variance in test scores accounted for by positional response bias, testwiseness, and guessing strategy was . 18 (R = .43, p < .05). Scores on the content exams were confounded by students' degree of testwiseness and, in six cases, by a positional response bias.