In order to increase customer satisfaction and competitiveness, manufacturing systems need to combine flexibility with Just-in-Time (JIT) production. Until now, research on JIT scheduling problems has been mostly limited to high volume assembly lines rather than job-shop-like systems, due to their combinatorial complexity. In this paper, we propose a generic strategy for dynamically controlling task schedules by coupling genetic algorithms and distributed arrival-time control to optimise JIT performance. We explore two such hybrid approaches: a sequential approach where the two algorithms work separately and an integrated approach where the distributed arrival time control is embedded into the genetic algorithm. Performance of these approaches is benchmarked with quadratic linear programme solutions to get a gauge of their relative strengths in a static environment. Results from applying these approaches to a job-shop-like automated cell verify their effectiveness for JIT manufacturing under realistic dynamically changing environment.