Abstract Rostering and staff scheduling problems often have a predefined set of shifts to which staff members can be allocated. These shifts are typically based on a small set of shift types, where each shift type is characterized by the period of the day the shift covers. The number of people assigned to a shift type reflects to some extent the demand for staff during the period it covers. We address the issue of whether or not we have the correct set of shift types given that we know the required staffing levels over a single day or a set of days. The Shift Design Problem (SDP) is the problem of identifying the set of shift types prior to the solution of rostering or staff scheduling problems. This should be done in such a way that the demand for staff in each period of the day is matched as closely as possible, but with restrictions on the number of the staff used each day as well as the number of shift types used. The SDP is a variant of the shift scheduling problem where -among other constraints -the number of shift types used is upper bounded. Shift scheduling has been considered by The motivation for the problem is based on the study by