International audience ; A survey based on the bacteriological analysis of 1328 honey samples from the harvest in the summer of 1999 yielded 146 samples contaminated with P. larvae spores (11.0% ), ranging from the detection of only a single colony of P. larvae to completely overgrown plates. Risk factor analysis by logistic regression modelling revealed that honey samples collected in the proximity of apiaries with AFB cases from 1998 were at 3-times higher risk of being contaminated compared to other geographic locations (odds ratio = 3.04; $P < 0.001$). Comb replacement was found to lower the risk (odds ratio = 0.46; $P = 0.033$) and should be encouraged in beekeeping management. There was also a relation between the positive honey samples collected in 1999 and the cases of AFB in the year 2000 (odds ratio = 2.64; $P = 0.009$), supporting the value of honey analysis for sanitary control of bee colonies.