We examined how population density, winter weather, snow conditions, and 2 large-scale climatic indices (North Atlantic Oscillation, NAO, and Arctic Oscillation, AO) influenced demography (reproduction and mortality) in an alpine herd of semi-domesticated reindeer Rangifer tarandus between 1959 and 2000 in Finnish Lapland. The herd lived on heavily grazed lichen pastures, with winter densities between 0.8 and 3.9 individuals km−2. Icing conditions occurred every 7th yr, on an average, and decreased reproductive rate (calves/females) by 49%. In general linear models icing remarkably increased the fit of snow models to reproductive rate. Incorporation of an interaction term between icing and the snow depth index provided better fit than a model without interaction. Delayed snowmelt decreased reproductive rate. For the day of snowmelt, however, the model without interaction was better than the interaction model. These 3 models provided the best fit to the data and accounted for 51–54% of the variation in reproductive rate. Winter mortality was related to density and large-scale climatic indices, but not to local winter weather except a slight increase in mortality during an icing winter. The best model for winter mortality, including reindeer density and NAO, accounted for 26% of variation in mortality. Three factors may be involved explaining weak density dependence or the lack of such dependence; climate change scenarios that predict higher winter temperature, more frequent thawing-freezing periods, and deeper snow would be expected to decrease reproductive rate and increase winter mortality of reindeer and thus to reduce profitability of reindeer husbandry. In contrast, early springs would be advantageous for reindeer in the short term.