The Oxford shoulder scores (OSS) and Constant scores were collected prospectively for 103 patients treated conservatively for proximal humeral fractures. The paired scores were analyzed by assessing the degree of correlation, the agreement between the scores, and their sensitivity to change. Linear regression analysis was also performed. The scores correlated well with a coefficient of .84 (P < .001), demonstrating a clear relationship between the scoring systems. There was good agreement between the scores and both were sensitive to change over time (P < .001). We have also produced a regression equation (R2 = .70, P < .001) to predict the Constant score from the OSS. This study documents and compares the behavior of two scoring systems commonly used in the assessment of patients with proximal humeral fractures. The OSS may be considered as an alternative for assessing longer term follow-up, as, being solely subjective, it is easier to administer and correlates well with the Constant score.