Expectations and requirements for wireless communication systems continue to grow rapidly. Reliability is one of the most challenging criterions. Soft combined turbo codes technique aims to improve the error correction capability of turbo codes by using soft combining principle, at bit level, inside the turbo decoder. In particular, this approach is based on weight factors which are key elements that impact the overall system's performance. In this paper, we evaluate the performance of soft combined turbo codes for Rayleigh flat fading channel and Max-Log-MAP decoding algorithm. Especially, we analyze the variation of the weight factors in order to obtain the associated optimal values that provide the minimum error rate. Simulation results show that the optimal weight factor associated to the extrinsic information previously stored increases as Energy per bit to Noise power spectral density ratio (Eb/No) decreases. Also, they demonstrate that soft combined turbo codes solution performs far better than the conventional scheme in fading environments.