We have previously reported in this journal that spatial ability influences academic performance in engineering. We have also reported that spatial ability is trainable, and can be increased through instruction focused on using perception and mental imagery in three-dimensional representation. In this article, we present the results of a longitudinal evaluative study of student difficulties with engineering graphics at our university, involving research over a 24-year period. Prior to our intervention, the first-year engineering graphics course at our university had a failure rate of 36% for all engineering students, and failure rates of 80% for African students studying at our university. In terms of outcomes, our study indicates that changed teaching practices have been associated with increased pass rates, which currently average 88% for all students. There is clear evidence that African students no longer experience major difficulties in passing the course. This can be attributed both to changes in teaching methodology, as well as broader contextual changes occurring in South Africa over the period reviewed. Despite evidence of increased pass rates, our data suggest that there is a continuing relationship between three-dimensional spatial perception and academic performance in engineering graphics, as well as with aggregated first-year engineering marks. For this reason, many students still require academic support. There is also evidence that female students experience difficulties with engineering graphics due to lack of previous technical experience, and use a variety of strategies to overcome their study problems. The most effective appear to be use of peers and friends, and in particular students who had previously taken the course, as mediators of the course content and materials. Those female students who have not taken technical drawing at school level, and female students who do not network with other engineering students, are those at greatest disadvantage. (Contains 5 tables.)