Underwater acoustic communications have received much attention in recent years leading to the development of powerful and reliable receiver algorithms for signal processing in a variety of ocean environments. With the feasibility of high-speed coherent underwater communications established, the focus of current research is shifting to more demanding communication scenarios, as encountered in underwater acoustic networks. The problem of signal detection in highly variant shallow water environments, in the presence of strong cochannel interference from other acoustic modems in the network is addressed. The signals of different users occupy the same bandwidth, making the respective channel responses and the underlying data symbols sequences the only distinction among distorted replicas observed at the receiver. The performance of a single-sensor decision-feedback equalizer (DFE) is compared to the performance of a centralized multiuser receiver that jointly performs adaptive equalization and interference cancellation, and a decentralized multi-sensor DFE that performs spatial signal combining and multichannel equalization. Performance at different receiver structures was tested on real data collected at a one-mile range in the Woods Hole harbor, from two closely separated sources. Excellent results were obtained at signal-to-interference ratios as low as −10 dB, showing the performance improvement of proposed techniques over DFE in a network scenario.