The use of adhesively bonded joints is often limited by a lack of reliable models able to accurately predict their behaviour in industrial applications, in which the stress distribution is often complex. The mechanical behaviour of an adhesive in a bonded joint is often heavily dependent on its stress state (i.e., the tensile-shear combinations). Thus, a large experimental database is required to accurately represent the complex behaviour of an adhesive in a bonded joint. On the one hand, the initial yield surface (initial elastic limit) often has to be described taking into account the two stress invariants, hydrostatic stress and von Mises equivalent stress, and on the other hand the non-linear behaviour of the adhesive is also quite complex to model. However, the mechanical response of adhesively bonded joints often presents quite large stress concentrations; thus, the analysis of experimental tests is made particularly difficult. Obtaining reliable experimental results makes it possible to contribute to optimization of an adhesive in a bonded joint. This paper presents comparisons between results of different experimental tests (with bulk and bonded joints), some of them are designed to greatly limit the edge effects. Results are presented for two adhesives under proportional monotonic loadings. The two adhesives have very different behaviours (a ductile adhesive and a brittle adhesive) and two different surface preparations of aluminium substrates (a mechanical preparation and a chemical preparation recommended by the adhesive manufacturer) were studied. (C) Koninklijke Brill NV, Leiden, 2011