Objective: To determine period prevalences of postmortem diagnoses for spinal cord or vertebral column lesions as underlying causes of ataxia (spinal ataxia) in horses., Animals: 2,861 client-owned horses (316 with ataxia [ataxic group] and 2,545 without ataxia [control group])., Procedures: The medical records database of the University of California-Davis Veterinary Medical Teaching Hospital was searched to identify horses necropsied between January 1, 2005, and December 31, 2017. Results were compared between the ataxic and control groups and between various groups of horses in the ataxic group. Period prevalences were determined for the most common causes of ataxia., Results: 2,861 horses underwent full necropsy, and the period prevalences for the most common definitive diagnoses for ataxia were 2.7% (77/2,861) for cervical vertebral compressive myelopathy (CVCM), 1.3% (38/2,861) for equine neuroaxonal dystrophy or equine degenerative myeloencephalopathy (eNAD-EDM), and 0.9% (25/2,861) for trauma; the period prevalence of ataxia of unknown origin was 2.0% (56/2,861). Horses in the ataxic group (vs the control group) were more likely to have been warmblood horses (OR, 2.70) and less likely to have been Arabian horses (OR, 0.53). In the ataxic group, horses < 5 (vs ≥ 5) years of age had greater odds of CVCM (OR, 2.82) or eNAD-EDM (OR, 6.17) versus trauma or ataxia of unknown origin. Horses in the ataxic group with CVCM were more likely Thoroughbreds (OR, 2.54), whereas horses with eNAD-EDM were more likely American Quarter Horses (OR, 2.95) and less likely Thoroughbreds (OR, 0.11)., Conclusions and Clinical Relevance: Results indicated that breed distributions differed for horses with CVCM versus eNAD-EDM; therefore, breed should be considered in the clinical evaluation of spinal ataxia in horses.