A nonlinear probabilistic model of the relaxation labeling (RL) process is implemented in the speaker identification task in order to disambiguate the labeling of the speech feature vectors. In this proposed algorithm, the deterministic labeling of the vector quantization (VQ)-based speaker identification is relaxed by means of introducing initial probabilistic weights to the labeling process of the speech feature vectors. This process is then iteratively updated until no further significant improvement is found. Experimental results on speaker identification using a commercial speech corpus show that the relaxation labeling outperforms the conventional VQ method.