Abstract There are only few studies that use both demographic and genetic data to assess population viability of plant species. We combined genetic and demographic data from 11 endangered perennial orchid populations of varying size in order to reveal determinants of viability. Small populations had substantially lower viability compared to large populations. Seedling recruitment rates were remarkably lower in small populations; this was not due to pollination limitation or inbreeding depression because the fruit set and heterozygosity were not correlated with population size, suggesting that there may be differences in successful germination. Low recruitment resulted in significantly lower predicted population growth rates in small populations. The impact of stochasticity on viability varied among populations and stochastic simulations indicated that only one large population was viable, whereas all the other large populations were predicted to go extinct within decades. While there was a positive correlation between the deterministic population growth rate and allelic richness, we did not find any other correlations between genetic variation and fitness or population size. The study populations are likely remnant populations of a once large meta-population that decreased in size due to unfavourable environmental conditions. Management should focus on the maintenance of large population size, which is needed to avoid negative consequences of stochasticity and to enhance seedling recruitment rates.