Nowadays, radar has been applied to human activity classification in the aging-in-place for health monitoring. The complex-valued neural networks (CVNNs) have been only minimally explored, especially on complex-valued radar signals, and there is an outstanding question on whether CVNNs can contribute to improving classification performance. This thesis proposes three complex-valued convolutional neural networks (CNNs) for human-motion classification based on monostatic radar. The range-time, range-Doppler, range-spectrum-time, and time-frequency spectrograms of micro-Doppler signatures are adopted as the input to CVNNs with different plural-handled approaches. A series of experiments determine the optimal approach and data format that achieves the highest classification accuracy. Experimental results on measured data show that 1) the accuracy of classification using CVNNs on range-Doppler and range-spectrum-time radar formats is significantly higher than the real-valued counterpart, and that 2)Deep neural networks achieve the best classification accuracy on CVNNs while shallow neural networks do not.