We have studied 16-QAM (quadrature amplitude modulation with four bits per symbol) modulation, a correlation receiver and a decision feedback equalizer (DFE), that together define a system for acoustic underwater communications. 16-QAM is a choice based on what has been reported in other studies and the experimental data that was given to us. At the receiver a decision feedback equalizer is used to handle intersymbol interference (ISI) due to multipath propagation. The main goal of this Master´s thesis is to evaluate parts of a system for underwater acoustic communication where channel models, modulation- and coding methods as well as receiving methods together should form a test bed in a MATLAB environment. We created baseband complex signals with 16-QAM modulation. These signals were passed through channel models constructed as finite impulse response (FIR) filters that mimic some effects of multipath propagation. Then noise was added to the signal, and the signal was then passed through a DFE. The DFE consists of a feedforward filter, a feedback filter and a correlation receiver. The optimization of the DFE was studied using both the least mean square (LMS) algorithm and solutions using the inverse covariance matrix of the training sequence. Without DFE the transmitted signal is totally corrupted and impossible to decode. But when the DFE is implemented, a large reduction of the error probability is achieved. For rather moderate ISI the symbol error probability is below 10-2 if the bit SNR is larger than 11 dB.