The workforce scheduling problem in call centres is defined as assigning a number of employees to various overlapping shifts during a specified planning horizon, considering some regulations and preferences, so as to minimize the labour cost. This is often difficult to solve to optimality due to its combinatorial structure. In this paper, we propose an enhanced artificial bee colony (EABC) algorithm to solve the problem. After introducing a good-quality solution generated by a heuristic algorithm to the initial population, the solutions are further exploited by four elaborate neighbourhood structures, which are designed based on the decomposed structure of the problem and deeply embedded in the solution evolution mode. Furthermore, a modified abandoning mode is employed in the onlooker stage. Using real instances from Chengxi Call Centre in Suqian, China, the experimental results show that the proposed algorithm can achieve (sub-)optimal solutions for large-scale problems. Furthermore, a comparative evaluation of EABC is carried out against a hybrid artificial bee colony (HABC) and simulated annealing (SA). The results show that EABC outperforms the other two algorithms for all problem instances. In addition, we analyse the influence of the weekend-off fairness on the labour cost.