Conventional chilled water based air conditioning systems use low temperature chilled water to remove both sensible load and latent load in conditioned space, and reheating devices are usually installed to warm the overcooled air, which leads to energy waste. Alternatively, this paper proposes a neural network (NN) model based predictive control strategy for simultaneous control of indoor air temperature and humidity by varying the speeds of compressor and supply air fan in a chilled water based air conditioning system. Firstly, a NN model has been developed to model the system dynamics, linking the variations of indoor air temperature and humidity with the variations of compressor speed and supply air fan speed. Subsequently, the NN model is experimentally validated and used as a predictor. Based on the NN model, a neural network predictive controller is proposed to control the indoor air temperature and humidity simultaneously. The experimental results demonstrate the effectiveness of the proposed scheme compared with conventional PID controllers. Moreover, it has been proven that it is practical to simultaneously control indoor air temperature and humidity by varying the compressor speed and the supply air fan speed without adding any other devices to the chilled water based air conditioning systems. (C) 2015 Elsevier B.V. All rights reserved.