American foulbrood (AFB) is a severe bacterial brood disease of honey bees (Apis mellifera L.) caused by the spore-forming bacterium Paenibacillus larvae. Low-levels of P. larvae in honey bee colonies are very widespread and can be considered endemic, whereas AFB occurs in clinical form only when a certain level of spores is reached in the colony. We studied for the first time the relationship between the level of the wintry contamination by P. larvae in materials taken from the hive and the onset of the disease in the following spring. In ten apiaries, from 125 colonies without signs of AFB, samples of adult bees, honey from the brood chamber and hive debris were collected in winter and cultured to detect P. larvae spore levels. The colonies were then checked in spring, until the end of May, for AFB symptoms. The performance of tests based on bees, honey and hive debris was evaluated, with respect to the results of clinical examination, by determination and comparison of sensitivity (Se), specificity (Sp) and positive predictive value (PPV) at different cut-off thresholds. Using the presence or absence of spores as a cut-off threshold, the examination of debris and bees showed a good Se (100%), while the Se of honey was limited (81.25%). At the same cut-off threshold, debris showed a greater Sp than that of bees (57.80% vs 7.34%). At increased cut-off thresholds, the Se was always higher for bee examination than for debris examination, whereas Sp and PPV were always higher for debris than for bees. The results showed that the levels of P. larvae spores in adult bees and, above all, in hive debris can be a useful tool for AFB risk assessment. The identification and proper management of the colonies with an increased risk of developing the disease is crucial for reducing the impact of AFB.