Abstract: Resource-constrained project scheduling problem (RCPSP) with activities splitable at fixed points is an extension of the standard RCPSP.We provide a new discrete artificial bee colony (DABC) algorithm for this problem.In DABC each food source location represents a feasible permutation of the set of activities.We give a new method to produce candidate food sources which can keep the solution discrete and feasible.The activities allowed to split at fixed points should be changed into subtasks which meet the first and last relations, and then the feasible scheduling is generated through the serial scheduling scheme.The actual computation shows that this algorithm can solve RCPSP with activities splitable at fixed points effectively.We also see that the project makespan can be shorten in a certain extent by splitting activities without affecting its quality.