The prevalence and number of salivary lactobacilli was determined in all schoolchildren starting the 7th grade (12-13 yr; n = 1,578) in comprehensive school located in Kumla, Sweden between 1987 and 1998. Whole saliva samples were collected and transferred to dip-slides (Dentocult-LB) and incubated in room temperature for 7 d. Caries data were collected from the dental records and from bitewing radiographs. Both the number of lactobacilli and the prevalence of manifest caries and restorations exhibited a decreasing tendency and were significantly lower in 1998 than in 1987. The mean caries prevalence (DFS) declined from 5.2 to 1.8. In 1987, 45% of the children harboured high or very high lactobacilli counts ( > or =10(5) CFU/ml) in their saliva compared to approximately 10-15% at the later examinations. The proportion of children with no detectable counts varied between 4-29% during the study period, while children with low counts (10(3)-10(4) CFU/ml saliva) constituted a clear majority at the recent samplings. With such low levels in the population, the use of lactobacilli counts as a didactic tool in dietary counselling must be called into question.