This paper questions the recommendation regarding the use of standard specimen geometries, (Type I, Type II, and Type IV), for estimating the tensile quasi-static and fatigue properties of structural epoxy adhesives. The work presents results from an experimental program investigating the performance of structural epoxy adhesives indicating a significant effect of the specimen geometry, especially when referring to fatigue loading. Simple finite element models are also developed to facilitate the comparison of the stress distribution along the three specimen geometries. The fatigue experimental results allowed the derivation of probabilistic S-N curves, showing higher fatigue sensitivity of Type I specimens compared to Type II and IV. Furthermore, probability distribution function (PDF) curves of the equivalent static strength estimated by using Sendeckyj's wear-out model attributed lower mean strength and higher variance for Type I specimens validating the fatigue data.