Background The Nordic Arthroplasty Register Association was initiated in 2007, and several papers about hip and knee arthroplasty have been published. Inspired by this, we aimed to examine the feasibility of merging data from the Nordic national shoulder arthroplasty registries by defining a common minimal data set. Methods A group of surgeons met in 2014 to discuss the feasibility of merging data from the national shoulder registries in Denmark, Norway, and Sweden. Differences in organization, definitions, variables, and outcome measures were discussed. A common minimal data set was defined as a set of variables containing only data that all registries could deliver and where consensus according to definition of the variables could be made. Results We agreed on a data set containing patient-related data (age, gender, and diagnosis), operative data (date, arthroplasty type and brand), and data in case of revision (date, reason for revision, and new arthroplasty brand). From 2004 to 2013, there were 19,857 primary arthroplasties reported. The most common indications were osteoarthritis (35%) and acute fracture (34%). The number of arthroplasties and especially the number of arthroplasties for osteoarthritis have increased in the study period. The most common arthroplasty type was total shoulder arthroplasty (34%) for osteoarthritis and stemmed hemiarthroplasty (90%) for acute fractures. Conclusion We were able to merge data from the Nordic national registries into 1 common data set; however, the set of details was reduced. We found considerable differences between the 3 countries regarding incidence of shoulder arthroplasty, age, diagnoses, and choice of arthroplasty type and brand.