The operative management of primary and secondary caries assumes that all discoloured tissue at the enamel-dentine junction (EDJ) represents active disease and this is removed to arrest the carious process. This study aims to establish clinical criteria to differentiate between active and arrested caries at the EDJ using microbiological assessment of dentine samples to verify its clinical status. Radiographs were available for posterior teeth. Cavities (n = 205) were prepared under rubber dam. After gaining access, areas of the EDJ were chosen and assessments made of consistency (soft, medium, hard), colour (dark brown, mid-brown, pale) and moisture content (wet, dry). Dentine was removed by using a No. 3 round burr and placed in 1 ml of bacteriological culture broth. This sampling procedure was repeated at the same site once during cavity preparation and again when the cavity was judged as fully prepared. Samples were vortexed, diluted and cultured to give viable counts of the total anaerobic microflora, mutans streptococci and lactobacilli; viable counts were expressed as log10 (CFU per sample +1). Results showed no significant differences between the microflora of primary and secondary caries. The number of bacteria recovered diminished significantly as cavities were completed. Initial samples from soft and wet lesions harboured significantly more bacteria, lactobacilli and mutans streptococci than samples from medium, hard or dry lesions. Lesions visible on radiographs harboured more bacteria, including lactobacilli and mutans streptococci, while dentine colour was not discriminatory.(ABSTRACT TRUNCATED AT 250 WORDS)