This study investigated the effectiveness of different combinations of examples and problems, including simultaneous examples and problems, example-problem pairs, and problem-example pairs on novices’ learning of optimization problems tasks in a calculus curriculum. Participants were college freshmen who major in business and management. Pretest-posttest control group design was used in this study. After pretest, learners were randomly assigned to three learning conditions Difficulty rating scale was measured in the learning phase when learning time was fixed. Results showed that in cognitive load measurement, learning from examples and problems simultaneously was more effective and efficient than example-problem pairs and problem-example pairs conditions. In posttest, there was no significant difference in effectiveness between learning from examples and problems simultaneously example-problem pairs conditions; however, they were both more effective than the problem-example pairs condition. Performance in far transfer tests had the same outcome as in near transfer tests among the three conditions. In addition, difficulty rating scale was sensitive to the measurement of cognitive load when learning time was fixed.