Risk factors associated with the occurrence of "neighbourhood infections" [Epidemiology of classical swine fever. In: Truszczynski, M. (Ed.), Proceedings of the Workshop on Diagnostic Procedures and Measures to Control Classical Swine Fever in Domestic Pigs and the European Wild Boar. Pulaway, Poland, pp. 119-130] during classical swine fever (CSF) outbreaks were examined based on information collected during a CSF-epidemic, which occurred in the East Flanders Province of Belgium in 1994. The only risk factor that was associated with the occurrence of "neighbourhood infections" was a kernel estimation of the intensity of neighbouring herds (P=0.055) [Interactive spatial data analysis. Pearson Education Limited, Harlow, Essex], i.e. the higher the kernel estimation, the higher the risk for the occurrence of neighbourhood infections. In a second part of the study, the likelihood for the occurrence of neighbourhood infections within an area with a 1 km radius was predicted for every Belgian pig herd, assuming that the herd was infected with CSF-virus. For the prediction of these likelihoods, the model resulting from the risk assessment was used. Finally, the predicted likelihoods were transformed into a raster map after applying a smoothing technique. As a result, different areas in Belgium of higher or lower risk for CSF-virus spread through "neighbourhood infections" could be identified on the map. The areas in Belgium where CSF-outbreaks including "neighbourhood infections" occurred in the past decades were all predicted by the model to be of high risk.