A serial concatenated convolutional and differential coding scheme is employed in a multi-user direct-sequence code-division multiple-access (DS-CDMA) system. The system consists of single-user detectors (SUDs), which are used to suppress multiple-access interference (MAI) with no requirement of other users' spreading codes, timing, or phase information. The differential code, treated as a convolutional code of code rate 1 and memory 1, does not sacrifice the coding efficiency and has the least number of states. The iterative process exchanges information between the differential decoder and the convolutional decoder. Both component decoders adopt the a posteriori probability (APP) algorithm. Numerical results in additive white Gaussian noise (AWGN) channels show that this concatenated coding scheme provides better performance and more flexibility than conventional convolutional codes in DS-CDMA systems, even in the sense of similar complexity. Further study shows that the performance of this coding scheme applied to DS-CDMA systems with SUDs improves by increasing the processing gain or the number of taps of the interference suppression filter, and degrades for higher near–far interfering power or additional near–far interfering users.