OBJECTIVE To describe the learning curve for veterinary surgery residents performing hemilaminectomy surgeries in dogs. DESIGN Retrospective case review and learning curve evaluation. SAMPLE 13 individuals who completed a 3-year surgery residency program at a university teaching hospital and who had no prior experience performing hemilaminectomies. PROCEDURES The 13 residents performed hemilaminectomies on 399 dogs between July 2006 and July 2013. Medical records were reviewed, and operative time was recorded. Data were examined with a linear mixed-effects model to quantify fixed and random effects, a curve-fitting technique to find the best-fit curve, and a segmented 2-phase linear model to describe the domains and learning rates for 2 phases of learning. RESULTS The linear mixed-effects model indicated that increasing patient body weight and increasing surgical complexity (graded on the basis of number and contiguity of hemilaminectomy sites) were associated with longer operative times and that increasing exposure number was associated with shorter operative times. The monoexponential and biexponential parametric curves were of similar quality in modeling the data. The segmented 2-phase linear model showed an early phase of learning during which operative time decreased rapidly and a late phase when operative time decreased more gradually. CONCLUSIONS AND CLINICAL RELEVANCE The learning curve for the residents suggested that for early exposures, instruction in the form of direct supervision provided substantial benefit. By the tenth exposure, the benefit of instruction diminished and ongoing improvement was primarily a result of refinement. If validated by further study, this understanding of a 2-phase learning curve may inform the design of training programs in veterinary surgery.