Abstract The distribution of runs of homozygosity (ROH) may be shaped by a number of interacting processes such as selection, recombination and population history, but little is known about the importance of these mechanisms in shaping ROH in wild populations. We combined an empirical dataset of >3,000 red deer genotyped at >35,000 genome-wide autosomal SNPs and evolutionary simulations to investigate the influence of each of these factors on ROH. We assessed ROH number and location in two populations of red deer (a focal and comparison) to investigate the effect of population history. We investigated the role of recombination using both a physical map and a genetic linkage map to search for ROH. We found differences in ROH distribution between both populations and both map types indicating that population history and local recombination rate have a substantial effect on ROH. Finally, we ran forward genetic simulations with varying population histories, recombination rates and levels of selection, allowing us to further interpret our empirical data. These simulations showed that population history has a greater effect on ROH distribution than either recombination or selection. We further show that selection can cause genomic regions where ROH are common, so called ROH hotspots, only when the effective population size (N e ) is large or selection is particularly strong. In populations having undergone a population bottleneck, genetic drift can outweigh the effect of selection. We show that most ROH hotspots in the Rum population are in line with expectations from neutral simulations, however two ROH hotspots show possible signatures of selection. Overall, we conclude that in this population, genetic drift resulting from a historical population bottleneck is most likely to have resulted in the observed ROH distribution, with selection possibly playing a minor role.