We assessed static skull variation in the Japanese weasel Mustela itatsi by integrating different variation indices. We used the coefficient of variation (CV), residuals of the standard deviation regressed onto the mean of each measurement (RSD) and allometry coefficients (ACs). CV showed nonlinear correlation with mean trait size as reported in many previous studies. RSD has a similar pattern of variation to CV and it has been used as an index to obliterate the trait size bias seen in CV. Furthermore, similarly sized traits showed a wide range of CV, with many largely scattered around a nonlinear regression curve. Therefore, variation in CV is not entirely generated by the bias and to some extent reflects real biological phenomena. Allometric analyses revealed that larger specimens tended to have a relatively larger viscerocranium, smaller neurocranium, more robust mandible, larger canines, smaller carnassials, and smaller M1 compared with smaller specimens. These patterns are possibly consistent with general ontogenetic skull shape change in the genus Mustela. Tooth measurements, as well as cranial and mandibular measurements, showed significant correlations with skull size. CV variation is determined mainly by AC and is weakly related to the correlation between trait and skull size.