This paper presents an evaluation of overstrength based on an experimental study on dowelled connections in cross-laminated timber (CLT). Connection overstrength needs to be well understood in order to ensure that ductile system behaviour and energy dissipation can be achieved under seismic loading. Overstrength is defined as the difference between the code-based strength, using characteristic material strengths, and the 95th 4 percentile of the true strength distribution. Many aspects contribute to total connection overstrength, which makes its definition challenging. In this study, half-hole embedment tests were performed on CLT to establish embedment strength properties and three point bending tests were performed to determine the fastener yield moment. Different connection layouts, making use of mild steel dowels and an internal steel plate, were tested under monotonic and cyclic loading to evaluate theoretically determined overstrength values and study the influence of cyclic loading on overstrength. Experimental results were compared with strength predictions from code provisions and analytical models for ductile response under monotonic loading. It was found that cyclic loading does not significantly influence overstrength for connections that respond in a mixed-mode ductile way indicating that in future more expedient monotonic test campaigns could be used. This work also provides further experimental data and theoretical considerations necessary for the estimation of a generally applicable overstrength factor for dowelled CLT connections.