Matched filter receivers are commonly used for Direct-Sequence Spread-Spectrum Multiple-Access communication systems. These receivers are easy to implement and analyze, and are optimal for single user Gaussian noise channels. However, for applications in spread-spectrum networks, lower average bit-error probability can be achieved by a receiver which takes into account the effect of the interfering users. Two such receivers which form Maximum-Likelihood decisions given an observation vector consisting of chip correlator outputs are shown to perform better than the matched filter. The performance of these receivers is analyzed via Monte Carlo simulations using Importance Sampling. The level of improvement over the matched filter is dependent upon the relative levels of Gaussian noise and multiple-access interference. These receivers demonstrate less deterioration of performance in near-far situations than the matched filter, and find application in wide-band radio networks.