Background: Spatial ability is significantly related to performance in engineering education, and problem-solving, an activity that is highly relevant to engineering education, has been linked to spatial ability. Purpose/Hypothesis: This study investigated the following question--To what extent is spatial ability related to problem-solving among engineering students and how do approaches to problem representation and solution vary with spatial ability level? Design/Method: A spatial ability test, a set of math word problems and an accompanying math ability test, was administered to two samples of first-year engineering students in two countries. Data were analyzed at the test level to evaluate the relationship of spatial ability to problem representation and solution. A detailed item level analysis was conducted to compare approach to problem-solving with spatial ability level. Results: Spatial ability was found to be significantly related to problem-solving but not to the math ability test, indicating the relationship was limited to the problem representation phase, not the solution phase. Key aspects of representation were identified for each problem to reveal a more pronounced relationship between representation and spatial ability than between problem-solving (representation and solution) and spatial ability. Conclusions: Problem-solving can be considered to consist of two cognitively distinct phases--problem representation and problem solution. It was found that spatial ability is significantly related to problem representation but not to problem solution. Hence, this study shows that spatial ability plays a key role in engineering education that is not limited to visualization of imagery but extends to thinking during problem-solving, a nonroutine activity that requires mental representation.