Objective The aim of this study was to determine the inter- and intra-observer variability of the modified Outerbridge cartilage classification system in canine joints evaluated via arthroscopy. Materials and Methods Fifty arthroscopic videos of canine cartilage were scored by six observers, where three of the observers had significant arthroscopic experience and three had minimal to no experience. The kappa (κ) statistic was used to evaluate overall and individual score inter-observer variability, as well as experience variability. The weighted κ statistic was used to evaluate the overall intra-observer variability for each observer, and for individual score intra-observer variability across experience groups. Landis and Koch cut-offs were used to determine strength of agreement associated with each κ-value. Results The overall inter-and intra-observer variability of the modified Outerbridge cartilage classification system showed fair and substantial strengths of agreement, respectively. The most extreme scores of 0 and 4 had the best inter- and intra-agreement and the middle scores of 1, 2 and 3 had decreased strengths of agreement. Experience of the observer increased the strength of agreement between the scores. Clinical Significance The modified Outerbridge classification system is an acceptable method for the evaluation of canine cartilage. Observer agreement is improved if the observer has experience with arthroscopy and viewing cartilage changes, and if the same observer is used for subsequent cartilage evaluations.