Samples of altered or carious dentin for microbiological culture were obtained from 301 primary root-caries lesions in 59 patients by means of a standardized sampling procedure. This involved the cleansing of each root surface of extraneous supragingival plaque by means of a hand-held toothbrush and distilled water and the collection of the sample with a sterile dental excavator passed through the entire vertical dimension of each lesion. The total number of colony-forming units (cfu) in each sample and the numbers of mutans streptococci (primarily Streptococcus mutans), lactobacilli, yeasts, and Gram-positive pleomorphic rods (GPPR) were determined. Individual bacterial counts were expressed as log10 (cfu per sample), as a percentage of the total number of bacteria per sample, and as a frequency of isolation from lesions with different clinical diagnostic criteria. Clinical measurements of each lesion were made for color, texture, position relative to the gingival margin, and treatment need. Lesions classified as soft yielded significantly more bacteria, mutans streptococci, lactobacilli, and GPPR than leathery lesions, which yielded more bacteria than hard lesions. Lesions were classified into 5 treatment categories: soft and restore, leathery and restore, leathery and debride of caries; leathery and treat therapeutically; and hard no treatment. The total numbers of bacteria, mutans streptococci, lactobacilli, GPPR, and yeasts decreased significantly with decreasing treatment need. The frequency of isolation of mutans streptococci, lactobacilli, and yeasts was significantly greater from lesions requiring restoration and from lesions situated within 1 mm of the gingival margin. We suggest that the majority of root caries is initiated adjacent to the gingival margin.