H.264 video standard is capable of providing good video quality at lower bit rates. Despite the low bit rates, H.264 video encoder suffers high computation complexity, resulting in substantial encoding time. In this paper, we propose a method based on JM source codes called the “Adaptive Mode Selection” (AMS) for motion estimation in H.264 video coding. The method can be presented in two phases: (1) Computation of adaptive values in reference frames; and (2) Adaptive mode selection. Our method was evaluated using 14 test videos and the results showed that the proposed AMS method, as compared with the JM full search method, could significantly save ~20% of encoding time, while maintaining low bit rates and video quality.