Annualised hours (possibility of irregularly distributing working hours over a year) permit companies to adapt capacity to demand, thus reducing overtime, temporary workers and inventory costs. To avoid a significant worsening of the working conditions, many laws and agreements constraint the distribution of working time. One way to constrain solutions is by specifying a finite set of weekly working hours and bounding the annual number of weeks of each type. Even though its impact on the solution, that set is usually agreed without considering all data available (demand, cost, etc.). In this paper two MILP models are used to determine in on step the most appropriate set of weekly working hours, the annual number of weeks of each type and the annual working time planning.