Studies concerning assessment of the image quality in scanning electron microscopes and studies evaluating the detective efficiency of the secondary electron (SE) detectors in these microscopes must be based on statistics of SE emission. The vast majority of previous studies have applied Poisson statistics, although their prerequisites have not been satisfied in most cases. This paper is concerned with the limits to the applicability of Poisson statistics to SE emission. Adequate definition of a non-Poisson factor in the variance of the number of SEs emitted is discussed, and a simple formula for this factor is derived for the low yield case in which both the primary and the backscattered electron are assumed not to release more than one SE. These conditions are met with conductive specimens composed of light elements at primary electron (PE) energies of tens of keV. For the lightest specimens, such as carbon, the non-Poisson factor can even be neglected for PEs >10 keV.