AbstractCarpentry joints are the traditional connecting media for timber elements. Among those, step joints are particularly important at structural size, because they are the most frequent solution to connect rafters and tie beams in roof timber trusses. The forthcoming version of Eurocode 5 will include, for the first time, provisions on its design. This paper describes the experimental investigation carried out to assess the safety and accuracy of the design formulae proposed in the literature and in some national standards on timber structures design. Although the behavior of double step joints is particularly aimed, given the scarce experimental information available on this configuration, single step joints were also investigated, intended to serve as a reference basis for the double-step case. It is also of interest to investigate how the load is shared between the two notches in the double-step geometries. To this end, an analytic procedure is proposed, based on the strains measured in the notches’ regions during the tests. Various test specimen groups were defined, whose dimensions favored one particular failure mode. The test setup was designed in order to simulate a rafter-to-tie beam connection in a timber truss. Besides the main strength tests of the step joints, side bending tests to EN 408 were also carried out in order to a better mechanical characterization of the used solid timber. Regarding the capacities, the results are in good agreement with the values provided by the formulae, and always on the safe side. The prevailing failure mode was as expected in most configurations, indicating the adequacy of the proposed design expressions. A procedure was proposed to determine the load share between the two notches in double step joints, but some setup detail changes are recommendable in order to improve its stability and reliability.