The effect of heat on the variability of lag phase of single cells was investigated. In addition, three alternatives on the lag phase determination were analysed, considering different approaches to estimate the inoculum size: samples with growth contained one initial cell (1C), the average of the truncated Poisson distribution of the number of cells (TP) or the number of cells given by the Poisson function (P). Increasing the intensity of the treatment lengthened the lag phase of survivors and increased its variability. In addition, the comparisons between the lag phase assuming the approaches 1C and P, and TP and P, usually differed significantly in the absence of heat treatment, but not when heat was applied. 1C and TP approaches were never significantly different. The assumption that growth starts from one cell is appropriate when lag phases are long, but it can lead to errors when growth conditions are optimal.