Most previous studies have focused on learners' self-efficacy in face-to-face learning environments, while few have delved into that of MOOC learners. No research has touched upon the MOOC self-efficacy of engineering students in developed and developing countries. This research compared the self-efficacy levels of engineering students in Taiwan and Vietnam and examined factors predicting their MOOC self-efficacy. An online 11-point Likert scale was sent to 222 students in two sites. An independent t-test was run to compare the MOOC self-efficacy levels of the students. A regression analysis model was used to understand which demographic variables were associated with the students' MOOC self-efficacy. Focus group discussions were conducted with 30 students. Quantitative and qualitative data helped to converge and corroborate research findings. There were no significant differences (p > 0.05) in the MOOC self-efficacy levels of the students in the two sites. Students in both universities gave the lowest ratings to the English self-efficacy subscale and the highest to the self-efficacy in the independent learning subscale. Mean scores ranged from 5.24 to 6.44. Students in Taiwan and Vietnam were moderately self-efficacious in three dimensions of the MOOC self-efficacy scale. Factors predicting self-efficacy included English proficiency levels, the number of prior MOOCs, age, and self-regulation. Implications for MOOC design and study were also given in the research.