There are many examples of buried rock surfaces whose age is of interest to geologists and archaeologists. Luminescence dating is a potential method which can be applied to dating such surfaces; as part of a research project which aims to develop such an approach, the degree of resetting of OSL signals in grains and slices from five different cobbles/boulders collected from a modern beach is investigated. All the rock surfaces are presumed to have been exposed to daylight for a prolonged period of time (weeks to years). Feldspar was identified as the preferred dosimeter because quartz extracts were insensitive. Dose recovery tests using solar simulator and IR diodes on both K-feldspar grains and solid slices taken from the inner parts of the rocks are discussed. Preheat plateau results using surface grains and slices show that significant thermal transfer in naturally bleached samples can be avoided by keeping preheat temperatures low. Equivalent doses from surface K-feldspar grains were highly scattered and much larger than expected (0.02 Gy to >100 Gy), while solid surface slices gave more reproducible small doses (mean = 0.17±0.02 Gy, n = 32). Neither crushing nor partial bleaching were found to be responsible for the large scattered doses from grains, nor did the inevitable contribution from Na-feldspar to the signal from solid slices explain the improved reproducibility in the slices. By modelling the increase of luminescence signal with distance into the rock surface, attenuation factors were derived for two samples. These indicate that, for instance, bleaching at a depth of 2 mm into these samples occurs at about ∼28% of the rate at the surface. We conclude that it should be possible to derive meaningful burial doses of >1 Gy from such cobbles; younger samples would probably require a correction for incomplete bleaching.