For labour intensive organizations, finding a good match between the predicted workload and the scheduled workforce work capacity is crucial: One important step in this matching process is shift design. In a multi-skill shift design problem, the model must reflect the skill attributes of the employees and the time-dependent demand of each skill type. For hospitals, a number of constraints and objectives complicate the picture, and in this paper we introduce models which reflect many of the challenges faced by planners at two reference hospitals in Norway. Experiments using mixed-integer programming solvers show promising results, and optimal solutions are in some cases found within seconds or a few minutes.