Phylogenetic relationships of several strains within the Fusarium oxysporum species complex and two related species from maize were used to test whether phylogenies inferred from nuclear-encoded translation elongation factor EF-1α and mitochondrial small subunit (mtSSU) ribosomal RNA gene sequences are concordant with one derived from random amplified polymorphic DNA (RAPD) characters. Results of a partition-homogeneity test (PHT) indicated that the EF-1α and the mtSSU rDNA data-sets are highly homogeneous (P ≥ 0.99) and could be analysed as a combined data-set. On the basis of the results of the PHT, the three RAPD partitions were also analysed as a combined data-set. However, the PHT revealed conflict between the nucleotide and RAPD data (P = 0.001). In addition, results of a t-test showed that the consistency, retention and rescaled consistency indices obtained with the nucleotide sequence data were significantly higher than those obtained with the RAPD data. The t-tests of the RAPD data also showed that these indices were significantly higher when using the 15-mer M13 RAPD primer compared with the two 10-mer primers. Although homoplasy was much greater for the RAPD data than for the DNA sequence data topologically similar trees were inferred from both data-sets. However the RAPD data appears to be flawed because it failed to resolve the two major clades of the F. oxysporum complex and trees constructed from the RAPD decamer primers completely failed to resolve the F. oxysporum complex. An unnamed phylogenetically distinct species (Fusarium sp.) intermediate between the F. oxysporum complex and Fusarium redolens was identified in trees inferred from the DNA sequence and RAPD data.