For practical reasons the numbers of mutans streptococci (MS) and lactobacilli (LB) in plaque are commonly estimated from saliva samples. The saliva counts are considered to be a reasonable indicator of the entire dentition's total microbial load. However, the value of salivary counts for explaining and predicting caries have been found to be low. There was therefore reason to compare the relationships between caries on the one hand and, on the other, the number of MS or LB in plaque and in saliva, respectively, in order to assess their relative merits for explaining the variation in caries, both in a total material and in subgroups with less favourable oral hygiene. Sixty children aged 14-15 years participated in the study. Caries and plaque were registered and the number of MS and LB was estimated in total plaque and in stimulated saliva samples. The results showed that the number of MS or LB in plaque did not explain the variation in caries to a greater degree than did the salivary counts. For both bacteria the explanatory values increased, as expected, in subgroups with less favourable oral hygiene, but not even at this higher level of explanation was there any difference between plaque and saliva. The LB count was a stronger explanatory variable than the MS count. It was concluded that the number of MS and LB, estimated in total plaque as well as in saliva samples, is not a useful tool in prediction.