Dike swarm emplacement accommodates extension during rifting and large igneous province (LIP) formation, with ancient dike swarms serving to localize strain during later tectonic events. Deciphering three-dimensional (3-D) dike swarm geometry is critical to accurately calculating magma volumes and magma-assisted crustal extension, allowing syn-emplacement mantle and tectonic processes to be interrogated. It is also important for quantifying the influence of ancient dike swarms on post-emplacement faulting. However, the essentially 2-D nature of Earth’s surface, combined with the difficulties in imaging subvertical dikes in seismic reflection data and the relatively low resolution of geophysical data in areas of active diking, means our understanding of dike swarm geometry at depth is limited. We examine an ∼25-km-wide, >100-km-long, west-southwest–trending dike swarm imaged, due to post-emplacement rotation to shallower dips, in high-quality 2-D and 3-D seismic reflection data offshore southern Norway. Tuned reflection packages correspond to thin (<75 m thick), closely spaced dikes. These data provide a unique opportunity to image and map an ancient dike swarm at variable structural levels. Crosscutting relationships indicate emplacement occurred in the Late Carboniferous–Early Permian, and was linked to the formation of the ca. 300 Ma Skagerrak-centered LIP. Dike swarm width increases with depth, suggesting that magma volume and crustal extension calculations based on surface exposures are dependent on the level of erosion. During the Mesozoic, rift-related faults localized above and exploited mechanical anisotropies within the dike swarm. We demonstrate that seismic reflection data are a powerful tool in understanding dike swarm geometry and the control of dikes on subsequent faulting.