When scrapie agent is exposed to partially inactivating autoclave cycles, the fraction of infectivity that survives remains thermostable during relatively long periods of autoclaving. This resistant subpopulation can also be differentiated from the main population by its prolonged incubation periods in assay animals, compared with control material. Stabilisation of this subpopulation may occur through the smearing and drying of infected tissue that can occur prior to autoclaving, in which the disease-specific form of PrP protein (PrP(Sc)) could become rapidly heat-fixed. This may paradoxically be what protects this fraction of PrP(Sc) from further inactivation during autoclaving. Data are presented showing that the thermostability acquired by the resistant subpopulation is a stable characteristic; autoclaving for a second time results in very little further loss of infectivity. These observations suggest that inactivation procedures that do not involve rapid and effective fixation of PrP(Sc) may be better candidates for dealing effectively with scrapie-like agents.