Self-tapping screws are fasteners that are versatilely applicable in timber engineering. For the design of such screw connections, preferential axial-loading, all possible failure mechanisms have to be considered. Recently, in compact groups of axially-loaded screws the block shear failure mechanism, which has not been investigated so far, turned out to fail rather brittle at load levels lower than currently allowed. This failure mechanism is defined as failure of (rolling) shear and/or tension perpendicular to grain planes encompassing the group of screws. This failure mechanism was observed in groups given a number of different parameter settings, i.e. thread-fibre angles of 90° and 45°, glulam, structural timber and cross laminated timber and various group designs. This paper focuses on groups of axially-loaded screws in glulam and solid timber of Norway spruce (Picea abies) and inserted at a thread-fibre angle of 90°. Varying group sizes, loading and supporting distances and group designs, i.e. various penetration lengths lef and spacing in and perpendicular to grain, a1 and a2, respectively, are analysed by two different “push-pull”-test setups. To predict the block shear capacity and failure characteristics of such groups of screws and to separate this failure mechanism from other failure mechanisms, a mechanical-based block shear model was established. This parallel acting spring model considers load sharing and redistribution between concerned failure planes and depends on a number of material, geometrical and stress distribution parameters. To ensure a reasonable parameter setting, background and potential influencing parameters on each model parameter are discussed. In validation, the model shows overall good predictions of capacities, failure mechanisms and failure sequence for all test series involved. It turned out that the current regulations, comprising the definition of minimum spacing together with minimum edge and end distances, are not sufficient for controlling this three-dimensional block shear failure. In addition, the consideration of the number of screws in the group as well as the penetration length is required.