To assess the epidemiology of Babesia divergens in a veterinary practice based in the mid-east of France ("Monts du Lyonnais"), blood was collected from 254 cattle belonging to 24 herds. To assess the dynamics of the carrier state, six carriers were identified, treated with flumethrin and sampled once every 3 weeks during 6 months. Two different DNA extraction methods were compared. Each sample was tested for the presence of parasites using a PCR-RFLP test based on the 18S rRNA gene. The sensitivity of the test was equivalent to a parasitaemia as low as 10(-5)% (in "Filter Paper" samples) and 10(-6)% in 1 ml blood (extracted using "Matrix"). With the latter method, the rate of detection diminishes in the low parasitaemia range but could probably be improved. This test proved to be very useful in the detection of B. divergens carriers. Serology using IFAT showed 7% of the cattle seropositive, which is suggestive of a disease situation with a low clinical risk level. Analysis of the PCR results suggests a 20% prevalence rate of carriers in the cattle population. The use of the mean parasitaemia is proposed to serve as a babesiosis clinical risk indicator. This approach could also be used in other babesia infections provided the lowest detectable parasitaemia level (threshold level) could be resolved for each parasite species.