This paper describes a method to diagnose the most frequent faults of a screw compressor and assess the magnitude of these faults. To determine the condition of the compressor, a feedforward neural network model is first identified from the compressor’s operating data. A recurrent neural network is then used to classify the model into one of three conditions including baseline, gaterotor wear and excessive friction. Finally, another recurrent neural network estimates the magnitude of fault from the model. The method’s ability to generalize was evaluated. Experimental validation of the method was also performed. The results show significant improvement over the previous method which used only feedforward neural networks.