Two maximum-likelihood (ML)-oriented data-aided (DA) symbol synchronisation algorithms are investigated in this paper. They operate on a noisy M-PSK signal (M > 2) transmitted over a slowly time-varying nonselective Rician fading channel. The first algorithm is channel-aided (CA): it makes use of an estimate of the instantaneous channel gain. The second algorithm is nonchannel-aided (NCA): timing is acquired independently of the channel gain estimation. Both algorithms operate either with a known training sequence or in a decision-directed mode, and are suitable for fully digital implementation. We And that the CA algorithm yields a smaller tracking error variance than the NCA algorithm, but the difference in performance decreases with increasing carrier-to-multipath (C/M) ratio.