Strength values of timber products, obtained by computer simulations, must be checked with regard to its absolute level and its apparent variation due to changes, inter alia, in the load configurations. In this paper, the focus is on the issue whether those apparent strength variations are reproduced correctly by the numerical procedure used. From physical perspective, apparent strength variations are caused by the interaction between the actual natural strength variability, changes in the load configuration and/or the member volume and the brittle material behaviour. The material used for this study is glulam modelled with brittle tensile failure. Finally, both the numerical and analytical procedure lead to almost identical results. Therefore, the comparison proves that the numerical approach more or less correctly reflects the basic principles of the Weibull theory since no contradictions between numerically and analytically determined strength variations exist. The study generally shows that corresponding analytical methods, if available, are a beneficial tool to verify and validate computational procedures.