Abstract: The reciprocating pump which is important equipment in the processes is widely used in drilling, water injection and fracturing. The working condition is severe, so the condition of monitoring and trend prediction of its wearing parts, such as pump valve, piston-cylinder liner and plunger-seal pair become the key problem to the safe operation of reciprocating pumps. This paper researches on fault trend prediction method of reciprocating pump based on gray-neural network, and evaluation of equipment condition using the prediction results. According to fault development trend and difficulties of fault diagnosis and prediction of reciprocating pumps, a combinative prediction model with grey and neural network is selected, which has higher prediction accuracy and the effective degrees of prediction.