Detrital zircon U–Pb geochronology is a rapidly expanding and useful technique for addressing the sedimentary rock record. However, because of difficulties in evaluating discordant analyses, common in most datasets, a high proportion of results may be discarded. These analyses, if interpreted correctly, can provide valuable information regarding the defor- mation, alteration or metamorphism of the zircon population, as these events are all capable of producing U–Pb discordance. A novel modelling procedure permits analysis of probabilistic relationships within U–Pb datasets to deconvolve the age information present within discordant analyses by assessing the relative likelihood of potential discordia lines. Additionally, because we retrieve useful information from discordant data, a stricter filter can be used to assess concordance, increasing confidence in those distributions. The validity of this modelling method is demonstrated using two previously published cases from the Caledonide orogen where clearly discordant analyses exist. In the Southern Uplands of Scotland, these analy- ses indicate resetting of the U–Pb systematics in metasedimentary rocks in the Grampian orogen prior to Ordovician erosion and redeposition. In the second case, from the Greenland Caledonides, discordant data provide additional constraints on previously proposed in situ resetting during Scandian deformation and fluid flow events.