We compared the congruence in genetic variation and phylogenetic relationships among mitochondrial DNA (mtDNA) haplotypes detected with restriction-fragment-length polymorphisms (RFLPs) and sequence analysis in the brook charr, Sulvelinus fontinalis Mitchill. This was accomplished by analyzing variation both at 172 restriction sites subsampling 903 bp over the entire mitochondrial genome and in the sequence of a 300-bp segment of the control region among the same 33 individuals. The sequence and RFLP analyses were highly congruent both in terms of mtDNA variants detected per number of nucleotides sampled and in assessing phylogenetic relationships among haplotypes. Therefore, variation observed in a small segment of the control region was representative of that existing over the entire genome. The major difference between both approaches was in the number of mutational changes observed per nucleotide in the sequence analysis. This number was approximately twice as high as that in the RFLP analysis. Nevertheless, the level of variation observed in the control region of charr was less than expected a priori on the basis of previous observations in other vertebrates. Lower mutation rates may be related to an unusually low transition: transversion ratio (8: 3) for mtDNA intraspecific variation in vertebrates. The combined restriction-site and sequence data resolved at least three distinct phylogenetic groupings in S. fontinalis. However, and despite the high-resolution level of the study (> 1,200 bp screened, or 7% of the mitochondrial genome), some major branching patterns in gene tree topology remained unresolved, which stresses the importance of recognizing large standard errors associated with sequence variability when phylogeny is reconstructed.