A new approach is proposed to calibrate and validate the most critical components of a dynamic traffic assignment (DTA) model: dynamic route choice and flow propagation. By presenting approximate joint probability distribution functions of the temporal link traffic flows on a network, it is possible to derive the likelihood functions for estimating dynamic route choice and actual flow propagation. The approach also enables statistical tests to be performed for validation of DTA models. Both procedures are presented with a small numerical example and a larger network. These examples also indicate that it is possible to calibrate and validate a DTA model with detection errors and incomplete data, especially when real-time traffic counts are available on only a few links in the network.