Deep unsupervised domain adaptation (Deep UDA) methods successfully leverage readily-accessible labeled source data to boost the performance on relevant but unlabeled target data. However, algorithm comparison is cumbersome in Deep UDA due to the lack of a satisfying and standardized model selection method, posing an obstacle to further advances in the field. Existing model selection methods for Deep UDA are either highly biased, constrained, unstable, or controversial (requiring labeled target data). To this end, we propose Deep Embedded Validation (\textbf{DEV}), which embeds adapted feature representation into the validation procedure to obtain unbiased target risk estimation with bounded variance. Variance is further reduced by the technique of control variate. The effectiveness of the proposed method is validated both theoretically and empirically.