Reported complications of neck dissection surgery have included decreases in shoulder muscle power and range of motion, drooping shoulder, and shoulder pain. We conducted a cross-sectional study to assess postoperative shoulder function following three different types of neck dissection surgery that were performed at Aga Khan University Hospital and to determine how various treatment factors and patient characteristics affected postoperative shoulder function. Our study population included 70 patients-51 men and 19 women, aged 18 to 70 years (mean: 48.6 ± 11.6)-who had undergone a total of 92 neck dissections (22 patients underwent bilateral procedures). Patients were assessed by physical examination and questionnaire responses. The physical examination included objective assessments of shoulder muscle power against resistance, active range of motion, maximum possible forward flexion, the length of time needed to repeat active shoulder flexion 5 times, and shoulder abduction. The questionnaire covered shoulder mobility during activities of daily living, the results of physiotherapy (and exercise), postoperative radiation status, and shoulder pain. Of the 92 neck dissections, 17 were selective (18.5%), 64 were modified radical (69.6%), and 11 were radical (12.0%). We found that patients who had undergone a nerve-sparing procedure (i.e., selective neck dissection or a modified radical neck dissection) exhibited significantly better shoulder function than did patients who had undergone radical neck dissections (p < 0.01). In addition, increasing age (p < 0.001) and a history of diabetes (p = 0.003) were associated with worse shoulder function, and postoperative physiotherapy was associated with better shoulder function (p = 0.002). Neither sex, weight, the side of the neck operated on (left or right), the administration of postoperative radiation, the length of time between surgery and shoulder function assessment, comorbidities such as hypertension and ischemic heart disease, nor the status of the level V lymph nodes had any statistically significant association with shoulder function.