Computer-based virtual learning environments (CBVLEs) are potentially useful teaching tools for training nursing students in professional duties such as the mathematical tasks associated with medication processes. In this study, a CBVLE was designed with well-structured instructional activities such as interleaved practice and feedback. Mathematical medication scenarios and basic arithmetic exercises were integrated into the CBVLE. Four training conditions were used in the CBVLE to facilitate extra support for mathematical medication learning: (1) learning without worked examples, (2) learning with worked examples involving domain-specific knowledge, (3) learning with worked examples involving regular thinking strategies, and (4) learning with combined worked examples. This study was conducted with 118 nursing students enrolled in post-secondary nursing education and Bachelor's nursing programmes. Students were pre-tested and post-tested on their mathematical medication learning. Training in the CBVLE improved mathematical medication learning for all students from pre-test to the post-test stages, but no differences were found among the four different conditions. Nursing students' prior knowledge, non-verbal intelligence, and number of correct tasks predicted mathematical medication learning outcomes. When controlling for non-verbal intelligence, students in the condition 1 benefited more than students in condition 3 in terms of their mathematical medication learning outcomes. The same accounted for the support of the low-achieving students in the CBVLE. The support conditions for the high-achieving group appeared to be unimportant for mathematical medication learning. It seems that technology is taken over some of the capacity of working memory, which accounts for the benefits to the low-achieving learners.