This work addresses the problem of channel intersymbol interference suppression in the presence of multiple sources. The approach, based on multiple-input/multiple-output (MIMO) decision-feedback equalization (DFE), extends a known single-user architecture that is well suited for operation in severely distortive channels. By working with a direct channel model to remove postcursor interference, the receiver can readily take advantage of known properties of the channel, such as sparsity of its impulse response, to improve the computational efficiency or tracking ability. Similarly to other MIMO equalizers, the finite-length coefficient matrices are optimized under an exact minimum mean-square error (MMSE) criterion. The MIMO setting offers additional flexibility relative to single-user scenarios that enables partial feedback of decisions pertaining to the current data block, in addition to previous blocks. Simulation results illustrate the performance of this DFE in single-carrier multiuser communication and single-user multicarrier modulation.