The leukocyte profile has the potential to be a reliable method to measure health conditions and stress in wild animals, but limitations occur because current knowledge on reference intervals is largely incomplete, especially because data come from studies on captive animals involving few individuals from single populations. Here we propose a general framework for achieving reliable leukocyte reference intervals, encompassing a set of internal and external factors, potentially affecting the leukogram. To do so, we present a systematic survey of the hematology of the common wall lizard, Podarcis muralis, involving 794 lizards from 54 populations over the whole geographic range of the species in Italy. Reference intervals for white blood cell (WBC) and leukocyte differential count were obtained by using linear mixed models in a Bayesian framework. The application of the procedure clearly showed that both internal (sex and size) and external (latitude and season) factors are a source of variation of leukocyte profile. Furthermore, the leukogram of common wall lizard has a strong variability among populations, which accounts for more than 50% of the whole variation. Consequently, some common assumptions used in studies on captive individuals are no longer supported in wild populations, namely, i) any group of individuals is a representative sample, ii) any population is representative of all others, iii) geographic clines do not occur over the species range, and iv) seasonal variation has limited effects. We encourage researchers aimed at the definition of leukocyte reference intervals for wild populations of reptiles to involve a large number of populations over a wide geographic range in ad hoc statistical models to disentangle local and geographic effects on leukocyte profile variation.