Background: There is consensus that most complete ulnar collateral ligament (UCL) injuries in throwers would benefit from surgical intervention. Optimal treatment for partial UCL tears remains controversial. Stress ultrasonography has become a well-accepted diagnostic modality for assessing UCL injury. Hypothesis: Partial UCL tears will result in an intermediate increase in ulnohumeral joint space gapping as compared with that of an intact UCL and a complete UCL tear, but the degree of joint space gapping will vary by anatomic location of the partial tear. Study Design: Controlled laboratory study. Methods: Twenty-one cadaveric elbows were divided into 7 groups representing different anatomic locations of UCL partial tears. Partial tears were simulated by cutting 50% of the measured width of the UCL at 6 locations: distal anterior/posterior, midsubstance anterior/posterior, and proximal anterior/posterior. A seventh partial tear was created by partially elevating the undersurface of the distal UCL to simulate the radiographic "T-sign." Valgus stress (15 daN) was applied to each cadaveric elbow at 30 degrees of flexion using a standardized device. Each specimen was tested intact, partially torn, and completely torn. At each state, joint space was measured using stress ultrasonography, and the difference in joint space from unstressed to stressed (delta) was recorded. Results: There were 10 right and 11 left upper extremity specimens. The mean delta was 0.58 mm for the intact state and <0.75 mm in all groups. Both distal partial tear groups had mean deltas <0.75 mm, similar to intact elbows. Proximal tears and the T-sign demonstrated intermediate deltas (0.99-1.23 mm). Midsubstance partial tears demonstrated the largest deltas (1.57-2.03 mm), similar to those of the complete tears. All complete tear groups had a mean delta >1.5 mm (1.54-3.03 mm). Conclusion: These findings suggest that partial tears introduce a spectrum of instability from functionally intact to completely torn. As a result, some may be biomechanically amenable to nonoperative treatment, while others would be indicated for early reconstruction. Further research into the biological and biomechanical determinants of nonoperative treatment failure will assist with more precise treatment recommendations.