The main aim of this study was to assess and compare undergraduate students' homework performance using a web-based testing system with paper-based, hand-graded one in introductory physics courses. Students' perceptions about each method were then investigated. Every semester during the two-year period, one of the two identical sections of introductory physics course students received a paper-based homework while the other a web-based homework. Then the two groups were compared on conceptual performance measures by using Force Concept Inventory Test in the first semesters and Conceptual Survey of Electricity and Magnetism Test in second semesters. Students in each group pre- and post- tested respectively each term by using a paper-based test. In general, statistically no significant differences were found in students' Force Concept Inventory and Conceptual Survey of Electricity and Magnetism test scores, while average homework performance scores varied each semester. Each year the homework performance difference for physics-1 was statistically significant in favor of paper-based homework group. However, homework performance difference scores for physics-2 were so varied each year that it cannot draw a conclusion about significance that could be attributed to the homework method used. Overall, it is discovered that students' perceptions about the web-based homework system were positive. (Contains 7 tables.)