Individuals tend to over or underestimate their knowledge and abilities. This study assessed the aptitude of students enrolled in beginning, intermediate, and advanced Animal Sciences courses to accurately predict performance on knowledge-based tests. Both easy and difficult knowledge domains were tested, and predication accuracy was determined following manipulation of test order and performance expectation. Actual performance was greatest for the easy knowledge domain. Presenting the easy test second or with a performance expectation of C- resulted in lower actual scores (P<0.05). Actual scores for the difficult test did not differ among the testing scenarios; however, students underestimated performance when the difficult test was completed second and overestimated performance when the difficult test was presented first with a performance expectation of B+ (P<0.05). Overall, prediction accuracy was improved for tests completed second. Participants were grouped into quartiles according to actual performance and quartile rankings were considered indicative of skill level. Students in the bottom and upper quartiles were most accurate in predicting performance on the easy test. Performance for the difficult test was overestimated when the difficult test occurred first with a performance expectation of C- and underestimated when completed second with a performance expectation of C-. Findings provide evidence that self-assessment of performance is influenced by task difficulty, task order, and performance expectation contributing to bias in self-reported data.