This study examines the efficacy of the Linear Congruent Method (LCM) in enhancing the integrity of Computer-Based Testing (CBT) systems through optimized question randomization. The primary objective was to evaluate how LCM can prevent question leakage and cheating by ensuring each test is unique and unpredictable. The research involved conducting 20 tests with different students, using both randomized and non-randomized settings to compare outcomes. Results indicated that LCM significantly improves test fairness and security by distributing questions in a manner that prevents predictability. Furthermore, this approach allows educators to accurately assess individual student understanding and tailor educational interventions effectively. Recommendations for future research include refining LCM parameters, integrating additional security technologies, and continuous evaluation to maximize the educational and integrity benefits of CBT systems.