A quantum self organizing feature map neural network (QSOM) method is introduced for water quality prediction in activated sludge wastewater treatment processes which includes uncertainty of microbial activity and complexity of biochemical reactions and strong lagging of parameters. This approach quantizes the inlet water quality data corresponding outlet water in abnormal state and makes the quantized data sample as the input of QSOM. The correlation coefficient of the quantum inputs and its weights are calculated as the best inputs matching of network by using quantum gates to update the weights in learning the rules. The experiments illustrate the efficiency of this prediction approach by using operational data of Chongqing Jiguanshi wastewater treatment plant.