The aim of this study was to compare measurement error (ME) between two different methods of measuring cranial traits: manual method, using calipers; and computerized one, using digitalized pictures and specialized software. Three observers measured 10 craniometric characters in 12 skulls of the common house mouse Mus musculus domesticus. Every measurement was repeated three times with each method. Nested ANOVA was used to separate the total variance into within- and among-individual components. Then the effect of trait size on ME was tested. Measuring method was the factor with higher values of ME, followed respectively by observer and replicate. Intra-observer variation was lower than inter-observer variation in both methods. However, repeatabilities were higher in the computerized procedure. Computerized measuring procedure was more precise and less influenced by factors increasing ME than manual method in most assessed traits.