Heat pump driven liquid desiccant dehumidification systems using hollow fiber membrane contactors have emerged as a promising and energy-efficient approach to air dehumidification. This paper reports on an optimization strategy for a two-stage system, which based on a set of heat and mass transfer models. The optimization problem was solved by a genetic algorithm to minimize the energy consumption. An experimental rig was built and experimental data were collected to validate the model-based operating strategy. The strategy implies that an hourly optimal regulation is proposed to control inlet temperatures on the inlet of membrane modules can be hourly regulated under hot and humid weather conditions. It appears that energy consumption is reduced by more than 20% to satisfy the indoor air humidity demand and the energy saving potential is more significant for the high moisture load. Slight variations of optimal energy input can stabilize electric power supply. Energy consumption of the compressor dominates the overall energy consumption and has the greatest potential to improve energy efficiency due to small differences between evaporating and condensing temperatures. The presented optimization strategy can be widely used as a real-time operation guide to monitor and control the systems under hot and humid weather conditions.