A new approach to model edge recombination in silicon solar cells is presented. The model accounts for recombination both at the edge of the quasi-neutral bulk as well as at an exposed space-charge-region (SCR), the latter via an edge-length-specific diode property with an ideality factor of 2: a localized J02, edge. The model is implemented in Quokka3, where the $J_{02,edge}$ is applied locally to the edges of the three-dimensional geometry, imposing less simplifying assumptions compared with the common way of applying it as an external diode. A “worst-case” value for $J_{02,{\rm{edge}}}$ , assuming very high surface recombination, is determined by fitting to full detailed device simulations which resolve the SCR recombination. A value of $\sim \text{19 nA/cm}$ is found, which is shown to be largely independent of device properties. The new approach is applied to model the impact of edge recombination on full cell performance for a substantial variety of device properties. It is found that recombination at the quasi-neutral bulk edge does not increase the $J_{02}$ of the dark J–V curve, but still shows a nonideal impact on the light J–V curve similar to the SCR recombination. This needs to be considered in the experimental evaluation of edge losses, which is commonly performed via fitting $J_{02}$ to dark J–V curves.