Limited research has been conducted on the spatial ability of veterinary students and how this is evaluated within anatomy assessments. This study describes the creation and evaluation of a split design multiple-choice question (MCQ) assessment (totaling 30 questions divided into 15 non-spatial MCQs and 15 spatial MCQs). Two cohorts were tested, one cohort received a 2D teaching method in the academic year 2014/15 (male = 15/108, female 93/108), and the second a 3D teaching method in the academic year 2015/16 (male 14/98, female 84/98). The evaluation of the MCQ demonstrated strong reliability (KR-20 = 0.71 2D and 0.63 3D) meaning the MCQ consistently tests the same construct. Factor analysis of the MCQ provides evidence of validity of the split design of the assessment (RR = 1.11, p = 0.013). Neither cohort outperformed on the non-spatial questions (p > 0.05), however, the 3D cohort performed statistically significantly higher on the spatial questions (p = 0.013). The results of this research support the design of a new anatomy assessment aimed at testing both anatomy knowledge and the problem-solving aspects of anatomical spatial ability. Furthermore, a 3D teaching method was shown to increase students' performance on anatomy questions testing spatial ability.