Shoulder bursitis and supraspinatus tendinitis are common conditions that result from the inflammation of the supraspinatus tendon. These cause pressure to be placed more heavily on the anterior bursa sac, along with the surrounding bones and nerves. A common treatment is conductive cooling on the affected region, generally in the form of a cold pack. However, if the cold pack is left on for too short of a time period, the cooling may not reach the tendon. The tendinitis would not be adequately treated in this case, as the inflammation could not be reduced if the tendon is not cooled to a temperature close to that of the rest of the body. If it is left on for too long, the patient may be subject to significant pain and the surrounding healthy tissue may be permanently damaged. Therefore, our goal is to identify the ideal treatment time for treating supraspinatus tendon inflammation. We created a three-dimensional model of the shoulder using COMSOL, with components integrated from Autodesk™ Computer aided design software and Google Sketch-Up. The implementation of our model into COMSOL allowed us to simulate the effects of this type of cold therapy on a human shoulder. A number of major parameter simplifications were required for adequate implementation into our model. Such simplifications included the grouping of the skin and muscle components, approximation of the humerus to a cylinder and a sphere, and grouping of the bursa sac and supraspinatus tendon into one domain because the two structures were too close together to be included in the model individually. We researched relevant literature to obtain property values for the bone and tissue components of our model. We used our best judgment to approximate the property values for parts of the geometry that were simplified, such as the tendon, muscle, and skin. In order to counteract any inaccuracy that could have resulted from imprecise averaged values, we performed sensitivity analysis to determine how tendon temperature varied with changes in the various material properties. This analysis supported our approximations. To further validate our model, we compared our results with previous research. The results of their study provided us with a guideline as to how much the muscle region should cool in a certain time period. Within that time period, our muscle tissue cooled by the expected amount, which validated our model. We decided that a 3-dimensional analysis of the supraspinatus tendon was necessary because the shoulder is asymmetric, so a 2-dimensional model would not be capable of capturing the necessary complexity. We determined that the standard suggested practice of treating an injury for a maximum of 20 minutes was not applicable for injuries that are as deep as the supraspinatus tendon. In this amount of time, the tendon’s temperature decreased by 3.05 Kelvin, when it needed to decrease by 7 K to be measured as successful cooling. We identified the ideal cold treatment time to be 96 minutes, using COMSOL to determine the point at which the supraspinatus tendon cooled within 1 K of body temperature. Further analysis indicated that the suprascapular nerve would not reach the threshold for cold pain of 288 K. However, other areas closer to the surface would reach temperatures well below this threshold. The cooling of the nerves running through this region could lead to significant pain. Literature shows that cooling for more than an hour could lead to permanent tissue damage, so 96 minutes of continuous cold treatment is not recommended. While our assumptions limit the clinical significance of this study, our results indicated that the use of cold therapy for only 20 minutes is ineffective because it does not reduce the temperature of the tendon enough to reduce inflammation. Cold treatment would likely be improved by the addition of anti-inflammatory drugs, in an attempt to combat the inflammation in two ways. We would recommend that further studies utilize unique heat transfer properties for the structures present in the shoulder instead of grouping them as one. This would bring the study closer to a clinically significant stage. In addition, further analysis of the likelihood of the pain response could be included by the addition of nerves to the model. Our model could also be used to test other cold or heat therapy technologies and their probable pervasiveness in the human shoulder, specifically the glenoid and sub-acromial regions.