Loss of genetic variation through genetic drift in isolated populations can reduce population viability. However, relatively little is known about loss of variation caused by fluctuating population size and realistic variance in reproductive success in age structured populations. We built an individual-based simulation model to examine how actual management strategies influence the effective population size (Ne) and allelic diversity in realistic population scenarios from Yellowstone bison. Simulation scenarios included a population size (Nc) ranging from 250 to >3,000 bison, a wide range of variance in male reproductive success, loci with 2, 5, or 20 alleles, and several realistic population culling strategies causing fluctuations in Nc. The resulting effective population size ranged from 746 to 1,176 with high to moderate variance in male reproductive success, respectively, for a stable population size of 2,000 bison. The Ne/Nc ratio remained stable around 0.33-0.38 for high variance in male reproductive success irrespective of the census size or the culling scenarios, but dropped to only 0.04 with extreme variance in male reproductive success. Heterozygosity was maintained at >95% over 200 years (~28 bison generations) for all simulation scenarios with Nc >500 and non-extreme variance in male reproductive success. The conservation of allelic diversity depended more on average Nc than Ne in fluctuating populations. Simulations suggest that 95% of allelic diversity will be maintained over 100 years if the Nc remains above 2,000 - 3,000 bison. However, less than 90% of alleles will be maintained at loci with more than five alleles (e.g. at immune system loci). Variance in male reproductive success had little effect on allelic diversity except under unrealistically-extreme variance in male reproductive success. These simulations are among the most realistic and extensive to date for a large mammal and should be informative in other species of large mammals with similar demography and mating systems.