OBJECTIVE To assess 3-D geometry of the humerus of dogs and determine whether the craniocaudal canal flare index (CFI) is associated with specific geometric features. SAMPLE CT images (n = 40) and radiographs (38) for 2 groups of skeletally mature nonchondrodystrophic dogs. PROCEDURES General dimensions (length, CFI, cortical thickness, and humeral head offset), curvature (shaft, humeral head, and glenoid cavity), version (humeral head and greater tubercle), and torsion were evaluated on CT images. Dogs were allocated into 3 groups on the basis of the craniocaudal CFI, and results were compared among these 3 groups. The CT measurements were compared with radiographic measurements obtained for another group of dogs. RESULTS Mean ± SD humeral head version was -75.9 ± 9.6° (range, -100.7° to -59.4°). Mean mechanical lateral distal humeral angle, mechanical caudal proximal humeral angle, and mechanical cranial distal humeral angle were 89.5 ± 3.5°, 50.2 ± 4.5°, and 72.9 ± 7.8°, respectively, and did not differ from corresponding radiographic measurements. Mean humeral curvature was 20.4 ± 4.4° (range, 9.6° to 30.5°). Mean craniocaudal CFI was 1.74 ± 0.18 (range, 1.37 to 2.10). Dogs with a high craniocaudal CFI had thicker cranial and medial cortices than dogs with a low craniocaudal CFI. Increased body weight was associated with a lower craniocaudal CFI. Radiographic and CT measurements of craniocaudal CFI and curvature differed significantly. CONCLUSIONS AND CLINICAL RELEVANCE CT-based 3-D reconstructions allowed the assessment of shaft angulation, torsion, and CFI. Radiographic and CT measurements of shaft curvature and CFI may differ.