Recently, many defect compensation methods are proposed for feed-forward neural networks implemented in hardware devices. However, there are few accurate quantitative comparisons with the performance of these defect compensation methods. In this paper,we compare the following three defect compensation methods;Partial Retrain- ing (PR)scheme,whole network back-propagation (BP)retraining and FT (Fault-T lerant)BP method.BP algorithm and PR scheme retrain the neural network after defects have occurred.FTBP method tries to obtain the weights those are robust for the defects. We can say that both BP algorithm and PR scheme are cure-type copensation methods and FTBP method cay say precaution-type compensation method. We compare the average recognition rate, average training time and the generalization ability among these three methods in detail. The experiments show that the whole network retraining by the BP algorithm has the highest reliability on the XOR problem and face image recognition problem on the neural networks with a single broken link defect and two broken link defects.