In an asynchronous direct-sequence code-division multiple access CDS-CDMA) communication system, the parameter estimation problem, i.e., estimating the propagation delay, attenuation and phase shift of each user's transmitted signal, may be complicated by the so-called near-far problem. The near-far problem occurs when the amplitudes of the users' received signals are very dissimilar, as the case might be in many important applications, In particular, the standard method for estimating the propagation delays will fail in a near-far situation, Several new estimators, the maximum likelihood, an approximative maximum likelihood and a subspace-based estimator, are therefore proposed and are shown to be robust against the near-far problem, No knowledge of the transmitted bits is assumed, and the proposed estimators can thus be used for both acquisition and tracking. In addition, the Cramer-Rao bound is derived for the parameter estimation problem.