Various methods have been proposed to estimate inbreeding depression and to assess its consequences for natural populations. As an alternative to controlled crosses, the use of molecular markers has allowed direct investigation of inbreeding depression in natural populations, but usually suffers from low statistical power. Here, we investigated the effect of inbreeding depression on survival in two populations of the rare species Brassica insularis, using both controlled crosses and a marker-based approach. We compare the respective merits of the two approaches for studying inbreeding depression. We also use information from the molecular markers to dissect in detail patterns of inbreeding depression in this species. A posteriori, we find that combining the approaches was not necessary to obtain simple point estimates of inbreeding depression. However, using molecular markers may give insight into the genetic basis of inbreeding depression, such as the occurrence of epistatic interactions among deleterious alleles or purging.