In 2017 part of an ancient stone causeway was uncovered at Karlslunde on the island of Sjælland, Denmark. With no artefacts found at the site, optically stimulated luminescence dating of coarse grains, grains derived from disaggregated rocks, and surficial rock chips obtained directly from granitic road cobbles were used to determine the time of construction. Some granitic road cobbles were visibly disaggregating at the time of excavation, and laboratory measurements revealed surprising strong fast component dominated quartz sensitivity from these samples. It was concluded that some of the rocks used in the causeway had been heated, presumably prior to incorporation in the structure. Dose recovery plateau experiments using sedimentary quartz and quartz grains recovered by gently disaggregating heated rocks suggested the use of a 220/180 °C preheat/cut-heat combination (DR ratio 0.999 ± 0.018; n = 40); this reduced the risk of thermal transfer in these young samples. IRSL signals were used for rocks that could not be disaggregated. The L/T burial profiles obtained from two unheated rocks (1 granite, 1 felsic gneiss) indicated they had been exposed for sufficient time for us to be confident of obtaining accurate IRSL ages. The post-IR IRSL180 signals were also measured in these two cobbles; the bleaching front was shallow and the signal was only sufficiently reset to allow accurate determination of De on one rock. After subtracting a residual, the dose recovery ratio results for the unheated rocks post-IR IRSL180 and IR50 were 1.006 ± 0.012 (n = 10) and 0.937 ± 0.007 (n = 10), respectively. In total 8 ages were accepted; 4 coarse grained sediment quartz ages from an unexcavated part of the road surface, 2 fading corrected IR50 ages from surface slices from non-disaggregated cobbles, and 2 quartz ages from disaggregated (apparently heated) cobbles. IRSL signals from the sedimentary and heated samples were used primarily to assess the degree of resetting of the quartz blue-stimulated OSL; both the post-IR IRSL and IR50 signals significantly over-estimated the quartz age. However, the sedimentary quartz ages, the heated cobble quartz ages and the fading corrected IR50 ages from the unheated road cobbles, are consistent and likely reflect a construction age of ∼2ka ago.