This study investigated the impact of computer-aided assessment on the enhancement of critical thinking skills in non-routine problem-solving across three distinct learning delivery modalities: fully online, balanced blended, and predominantly face-to-face (F2F) blended. The study aimed to evaluate the effectiveness of computer-aided assessment in enhancing students' critical thinking skills through non-routine problem-solving across three distinct learning delivery modalities. A quasi-experimental non-equivalent pre-test and post-test design was employed. Participants were drawn from three higher education institutions, each offering one of the three learning delivery modalities. In each institution, two intact groups were conveniently chosen and then randomly assigned to an experimental group that engaged with computer-aided assessment and a control group that used printed or digital problem sets. Data were analyzed using mean and standard deviation, paired and independent t-test and two-way ANOVA. Results showed that computer-aided assessment enhanced students' critical thinking skills through non-routine problem-solving, with varying effects across different learning delivery modalities. The balanced blended modality benefited the most from the computer-aided assessment intervention and proved most effective, showing the highest and most consistent gains, while the face-to-face blended and fully online modalities showed different patterns of improvement. These results suggest the importance of tailoring instructional strategies to specific learning environments to maximize the impact of computer-aided assessment.