The degree of binding was determined between deoxyribonucleic acid (DNA) preparations from nocardiae and "rhodochrous" strains and uracil-labeled DNA from three reference strains, Nocardia asteroides N668 and rhodochrous strains N11 and N54. In all cases, good congruence was found between the DNA reassociation data and that from numerical phenetic studies. Only a small degree of nucleotide sequence homology was found between the N . asteroides reference system and the other taxa studied, and there was evidence that N .crsteroides is genetically heterogeneous. The moles percent guanine plus cytosine for the rhodochrous strains was within the range 58 to 67; the corresponding range for nocardiae was 64 to 68 mol%.The confused and tortuous taxonomic history of bacteria variously known as "Mycobacterium" rhodochrous, the "rhodochrous" complex, and the "rhodochrous" taxon was reviewed by Bousfield and Goodfellow (51, who concluded that these organisms form a recognisable taxon equivalent in rank to the genera Corynebacterium, Nocardia, and Mycobacter i u m . The internal structure of the rhodochrous complex is still obscure, but the results of chemical (2, 3, 17), numerical phenetic (17, 39), deoxyribonucleic acid (DNA) reassociation (lo), genetic recombination (l), and serological (24) studies show that subgroups exist.The heterogeneity of the rhodochrous complex is most clearly seen in numerical phenetic studies in which two or more homogeneous phena have been recognised (6,16,17,22,35). However, since few strains are common to all of these investigations, it is difficult to determine whether or not, and to what extent, the phena overlap. In an extensive numerical taxonomic study, Goodfellow and Alderson (J. Gen. Microbiol., in press) divided 150 representative rhodochrous strains into 10 homogeneous subclusters (1A through 15). Subcluster 1A was equated with N . rubra (6), Gordona rhodochroa (38)) and phenon la (17); subclusters 1B and 1E were equated with phena 14B and 14A (16), respectively; subcluster 1D was equated with N . pellegrino (28); subclusters 1F and 1G were equated with phena lc and lb (17), respectively; subcluster 1H was equated with N . erythropolis (6) and phena 14D (16) and F3 (22); and subcluster 1J was equated with bacteria given the trivial name "Lspi" (large, spored, pink, irregular) (15). Subcluster 1C can be subdivided further and contains strains of G. bronchialis, G. rubra, and G . terrae (37).Good congruence has been observed between nucleotide sequence homology and numerical phenetic data in a number of genera including actinomycete taxa (13,19,34). In most cases, taxospecies share at least 70% DNA homology, and lower binding values are considered to reflect significant genetic divergence (12). In an earlier DNA reassociation study, Mordarski et al. (29) found that representatives of G . bronchialis, G . rubra, G . terrae, N . pellegrino (28), and phenon la (17) formed DNAhomology groups. In view of these encouraging results, we have extended our DNA reassociation assays on nocardiofo...