In this paper, we consider a task assignment problem arising in an assembly shop, in which n tasks are assigned to m workers. The integer processing time Pi[j] is required when worker i processes task j. For an assignment of the n tasks, the load of each worker means the sum of processing times of tasks assigned to the worker. Since every worker processes the assigned tasks repeatedly during a day, the maximum load over all the m workers is referred to as the cycle time. The minimization of the cycle time is closely related to the maximization of throughput of the assembly shop. The objective is to find an assignment of tasks that minimizes the cycle time under some specific constraints, but the problem is NP-hard. In this paper, we propose a heuristic algorithm using a binary search technique for the task assignment problem, which runs in polynomial time. The performance is examined by means of numerical experiments, and the results are reported.