Microbial thermal inactivation survivor curves (log10 numbers plotted against time) have long been described as maintaining a strictly linear rate of decline. However, much evidence exists which suggests deviation from log-linear kinetics does occur, and that this is not purely the result of experimental procedure as contended by some authors. Here, the shape of inactivation kinetics in Salmonella enteritidis was investigated. A heat challenge method was developed which, as far as could be ascertained, was free from methodological artefacts influencing the shape of survivor curves. High initial cell densities allied to sensitive enumeration resulted in biphasic survivor curves at 60°C. Tailing survivors accounted for approximately 1 in 105 of the initial population and possessed roughly four times the heat resistance. At temperatures 50 to 65°C, the presence of tailing prevented the use of D-values to accurately describe death rates. However, describing survivor curves using a log-logistic model increased data-fit at all temperatures investigated. The biphasic nature of survivor curves was studied closely between 49 and 60°C. It was observed that the extent of tailing was temperature-dependent; as temperature decreased, linearity increased such that at 51°C, survivor curves had no tailing. Studies using S. typhimurium and S. senftenberg 775W revealed similar kinetics. In these salmonellas, survivor curves demonstrated linearity at 54 and 57°C, respectively. The influences of culture age and growth rate on the shape of 60°C-inactivation curves were also investigated. Batch-cultured S. enteritidis cells of various maturities gave rise to survivor curves of differing heat sensitivities. Exponentially growing cells were shown to be the most heat sensitive, while late-stationary phase cells were the only populations to result in non-tailed survivor curves. Carbon-limited continuously cultivated cells demonstrated similar biphasic inactivation kinetics. Predictably, the slowest dilution rate corresponded to the greatest heat resistance. Starved cells produced linear inactivation kinetics that were virtually identical to those of late-stationary phase batch-cultured cells. That tailing in batch cultures was similar to chemostat populations, indicated that possible differences in growth rates in batch-cultured cells could not account for tailing. Furthermore, growth was necessary for tailing to be observed. Investigations into the cause of tailing revealed that these cells were not genetically distinct from the majority population. Instead, it is believed that tailing cells arise following the expression of heat-shock proteins during heating. Partial inhibition of de novo protein synthesis during heating resulted in much reduced levels of tailing. It is proposed that the temperature of inactivation determines the proportion of cells capable of expressing a heat-shock response, such that the temperature at which linearity is achieved corresponds to the point at which all cells are fully heat-shock protected.