Abstract Mineralogical and geochemical data from shallow overburden surveys are examined to ascertain parameters which govern the distribution of gold in overburden in a 2400-km 2 area of southwestern Gaspesie, Quebec, Canada. The area is a deeply dissected plateau underlain by faulted and gently folded Siluro-Devonian strata. Complex geomorphic and glacial histories are reflected in the non-glacial character of the landscape, the preservation of very old erosional landforms and extensive variation in the composition and distribution of overburden. Total sample analysis and heavy-mineral studies show that the composition of overburden changes across the area in approximate correspondence with changes in underlying bedrock. Three broad zones related to bedrock and overburden types are delimited. Gold analyses of μ m overburden are insensitive to regional variations, with only 15 samples out of 300 registering above the detection limit of 2 ppb. Better contrasts of gold concentrations are obtained from chemical analyses of nonmagnetic heavy-mineral concentrates (NM HMC). Although sample density is low, NM HMC data show anomalies which can be related to particular bedrock and structural settings. Particulate gold was not observed in any of the NM HMC. Gold is associated with secondary iron-oxide phases replacing primary sulphide minerals. High concentrations of gold in NM HMC of overburden collected north of the mouth of the Assemetquagan River support the hypothesis of a local source to the north or northwest for the alluvial gold in the lower 2 km segment of the river. Dilution by far-travelled, shield-derived heavy minerals is by far the most important cause of regional mineral variation. Conversion of gold concentrations in NM HMC to concentrations in total size fraction eliminates some of the erratic behavior of NM HMC data caused by variations in heavy-mineral abundance and corrects for the dilution effect where the proportion of heavy minerals in the far-travelled component of overburden is much greater than in the local component. These calculations suggest a contribution of gold to the background in the fine sand fraction of overburden of 0.07 ppb. Where the diluting component is local, conversion of NM HMC data to total size fraction may or may not correct for differences in heavy-mineral contributions of underlying bedrock, depending on local conditions. Where dilution by far-travelled components is excessive, NM HMC analyses are inadequate to reflect conditions in underlying bedrock.