Because of the confounding effects of long incubation duration and flock management, accurate epidemiological studies of scrapie outbreaks are difficult to carry out. In this study, 641 Manech red-faced sheep from six scrapie-affected field flocks in Pyrénées Atlantiques, France, were monitored for clinical scrapie over a 6-9 year period. Over this period, 170 scrapie clinical cases were recorded and half of the culled animals were submitted for post-mortem transmissible spongiform encephalopathy diagnosis to assess their infectious status. Collected data were analysed using a 'mixture cure model' approach, which allowed for the discriminating effect of PrP genotype and flock origin on incidence and incubation period. Simulations were performed to evaluate the applicability of such a statistical model to the collected data. As expected, ARR heterozygote sheep were less at risk of becoming infected than ARQ/ARQ individuals and had a greater age at clinical onset. Conversely, when compared with ARQ/ARQ, the VRQ haplotype was associated with an increased infection risk, but not a shorter incubation period. Considering the flock effect, we observed that a high incidence rate was not associated with shorter incubation periods and that the incubation period could be significantly different in flocks harbouring similar infection risks. These results strongly support the conclusion that other parameters, such as the nature of the agent or flock management, could interfere with epidemiological dynamics of the infection in scrapie-affected flocks.