This paper considers the issue of spreading code allocation and receiver optimization in a CDMA-based multiuser data network. Assuming that non-linear serial interference cancellation is performed at the receiver, the problem of noncooperative SINR maximization with respect to each user's spreading code and uplink receiver is first considered. It is shown that iterative SINR maximization converges in a finite number of steps, and that the equilibrium resulting from noncooperative behavior is also Pareto-optimal. Next, the problem of optimal multiuser detection for a system using the above optimal spreading codes is also considered, and it is shown that, for such a system, maximum likelihood detection entails a computational complexity that is exponential in the ratio between the number of users and the system processing gain, thus providing huge savings with respect to the general case wherein the optimal receiver has a complexity exponential in the number of users. While initially considering a simple frequencyflat synchronous CDMA system, the obtained results are then extended to more general scenarios, such as asynchronous CDMA systems, multipath CDMA channels, and multiuser multiantenna systems. Finally, the issue of non-cooperative power minimization and choice of the spreading code subject to a QoS constraints is also considered, and comparisons with the corresponding cooperative solution are also discussed. Numerical results corroborate the validity of the theoretical analysis, and show that the use of non-linear interference cancellation coupled with waveform adaptation brings remarkable performance gains with respect to linear multiuser receivers.