Multiple group principal component analysis was used to investigate the morphological relationships between feral American mink from three areas of Southern Norway based on 15 skull dimensions. The first principal axis calculated from the variance-covariance matrix of log-transformed data is usually interpreted as a growth axis. In the present analysis, the scores of the first principal axis of males were found to be uncorrelated with estimated age for specimens younger than seven months old, and was interpreted as a growth-free size axis. However, the second component of males was found to reflect mainly growth in the postorbital area. The first component of females was thought to reflect a mixing of growth-out and growth-in size variation. One of the reasons for separating size and size-free variation in the present work was to test whether growth-dependent size influenced the assessment of geographic variation within each sex. The scores on the components reflecting size variation (growthout and growth-in) were, however, not found to deviate significantly between the areas. Discriminant analyses run on the component scores, both with size variation included and excluded, gave the same result. In mean, the mink from Western Norway deviated significantly from the mink from Trondelag and Eastern Norway, while the mink from the two latter areas did not deviate morphometrically. The deviation found was in shape rather than in size of both sexes.