Aims. Our aim was to investigate the prevalence of Propionibacterium (P.) acnes in the subcutaneous fat and capsule of patients undergoing shoulder surgery for frozen shoulder or instability. Patients and Methods. A total of 46 patients undergoing either an arthroscopic capsular release or stabilisation had biopsies taken from the subcutaneous fat and capsule of the shoulder at the time of surgery. These samples were sent for culture in enrichment, and also for Nucleic Acid Amplification testing. The prevalence of P. acnes and other microbes was recorded. Fisher's exact test of binary variables was used to calculate the association with significance set at p < 0.05. Assessment of influence of independent variables including a pre-operative glenohumeral injection, fat colonisation and gender, was undertaken using binary linear regression. Results. A total of 25 patients (53%) had P. acnes in one or more tissue samples and 35 (74%) had other bacterial species. The same microbe was found in the subcutaneous fat and the capsule in 13 patients (28%). There was no statistically significant association between the surgical pathology and capsular colonisation with P. acnes (p = 0.18) or mixed identified bacterial species (p = 0.77). Male gender was significantly associated with an increased capsular colonisation of P. acnes (odds ratio (OR) 12.38, 95% confidence interval (CI) 1.43 to 106.77, p = 0.02). A pre-operative glenohumeral injection was significantly associated with capsular P. acnes colonisation (OR 5.63, 95% CI 1.07 to 29.61, p = 0.04. Positive fat colonisation with P. acnes was significantly associated with capsular P. acnes (OR 363, 95% CI 20.90 to 6304.19, p < 0.01). Regression models pseudo R. 2. found fat colonisation with P. acnes to explain 70% of the variance of the model. Patients who had a pre-operative glenohumeral injection who were found intra-operatively to have fat colonisation with P. acnes had a statistically significant association with colonisation of their capsule with P. acnes (OR 165, 95% CI 13.51 to 2015.24, p < 0.01). Conclusion. These results show a statistically significant association between subcutaneous skin P. acnes culture and P. acnes capsular culture, especially when the patient has undergone a previous injection. The results refute the hypothesis that P. acnes causes frozen shoulder. Cite this article: Bone Joint J 2017;99-B:1067–72