The maximum likelihood sequence estimator for the reception of coded digital phase modulated signals with single or multiamplitude constellations, transmitted over a multiplicative, frequency-nonselective (i.e., flat) correlated fast fading Rayleigh or Rician channel and corrupted by additive white Gaussian noise (AWGN), is derived. Due to this correlation, the errors caused by fading tend to occur in bursts. In our analysis, no assumption simplifying the problem is made, for example, breaking up these error bursts by means of interleaving/deinterleaving. For fast fading we consider the most general case where both phase and amplitude distortion resulting from the fading process could change significantly and thus cannot be assumed to be constant over a number of transmitted symbols. It is shown that the estimator's hardware structure consists of a combination of envelope, multiple differential and coherent detectors. With multiple differential detectors we define a receiver structure consisting of a combination of more than one distinct differential detectors each of them employing a progressively increasing (by the symbol duration) time-delay element (see Fig. 6). The outputs of these detectors ace jointly processed by means of an algorithm which is presented in a recursive form. The derivation of this new receiver is general enough to accommodate trellis coded phase shift keying (PSK) and quadrature amplitude modulated (QAM) systems. Differentially encoded signals, such as the pi/4-shift differential quadrature phase shift keying (DQPSK) scheme can also be incorporated. In order to reduce the overall receiver implementation complexity, several reduced complexity, near-optimal versions of the algorithm are presented. These reduced complexity receivers are based on the use of only a Pew multiple differential detectors. Performance evaluation results for reduced complexity trellis coded pi/4-shift DQPSE;, pi/4-shift 8-DQAM (differential quadrature amplitude modulation) and 8-DPSK (differential phase shift keying) systems have demonstrated that the proposed receivers significantly reduce the error floors caused by fading. What is even more interesting though, is that this performance was achieved by employing only a few multiple differential detectors, thus maintaining reasonable levels of implementation complexity.