In this paper, we describe the transmission of Classical Swine Fever virus (CSF virus) within herds during the 1997-1998 epidemic in The Netherlands. In seven herds where the infection started among individually housed breeding stock, all breeding pigs had been tested for antibodies to CSF virus shortly before depopulation. Based upon these data, the transmission of CSF virus between pigs was described as exponential growth in time with a parameter r, that was estimated at 0.108 (95% confidence interval (95% CI) 0.060-0.156). The accompanying per-generation transmission (expressed as the basic reproduction ratio, R0) was estimated at 2.9. Based upon this characterisation, a calculation method was derived with which serological findings at depopulation can be used to calculate the period in which the virus was with a certain probability introduced into that breeding stock. This model was used to estimate the period when the virus had been introduced into 34 herds where the infection started in the breeding section. Of these herds, only a single contact with a herd previously infected had been traced. However, in contrast with the seven previously mentioned herds, only a sample of the breeding pigs had been tested before depopulation (as was the common procedure during the epidemic). The observed number of days between the single contact with an infected herd and the day of sampling of these 34 herds fitted well in the model. Thus, we concluded that the model and transmission parameter was in agreement with the transmission between breeding pigs in these herds. Because of the limited sample size and because it was usually unknown in which specific pen the infection started, we were unable to estimate transmission parameters for weaned piglets and finishing pigs from the data collected during the epidemic. However, from the results of controlled experiments in which R0 was estimated as 81 between weaned piglets and 14 between heavy finishing pigs (Laevens et al., 1998a. Vet. Quart. 20, 41-45; Laevens et al., 1999. Ph.D. Thesis), we constructed a simple model to describe the transmission of CSF virus in compartments (rooms) housing finishing pigs and weaned piglets. From the number of pens per compartment, the number of pigs per pen, the numbers of pigs tested for antibodies to CSF virus and the distribution of the seropositive pigs in the compartment, this model gives again a period in which the virus most probably entered the herd. Using the findings in 41 herds where the infection started in the section of the finishers or weaned piglets of the age of 8 weeks or older, and of which only a single contact with a herd previously infected was known, there was no reason to reject the model. Thus, we concluded that the transmission between weaned piglets and finishing pigs during the epidemic was not significantly different from the transmission observed in the experiments.