Iterative equalization using optimal Multi-User Detector (MUD) & trellis-based channel decoder in coded Code Division Multiple Access (CDMA) systems improves Bit Error Rate (BER) performance. When large number of users is employed in the system over multipath channels, causing significant Multiple-Access Interference (MAI) & Inter Symbol Interference (ISI), the optimal MUD is thus complex. Therefore, the sub-optimal detectors such as low-complexity linear & nonlinear equalizers have to be considered. In this paper, Recursive Least Square (RLS) adaptation algorithm for Adaptive Decision Feedback Equalizer (ADFE) is proposed. Along with the application of the adaptive method to the DFE-RLS equalizer, Turbo-principle can easily be applied. Based on the conventional DFE, a new method is developed by computing the reliable extrinsic Log-Likelihood Ratio (LLR). At each iteration, the estimated symbol by the equalizer is then saved as a priori information for next iteration. Simulation results demonstrate that proposed DFE-RLS algorithm & turbo decoding, offers performance gain improvement of 0.7dB over DFE-LMS.