How can conservation planners optimally and effectively allocate limited resources between imminently threatened and presently secure areas? Such choices must be made at multiple spatial scales involving a variety of conservation targets. Allocation strategies range from a “fire-fighting” approach, which gives priority to heavily developed areas at high risk of further habitat loss, to a “pre-emptive” approach giving priority to intact habitat tracts before they become threatened. We determined optimal dynamic reserve selection strategies when selections are made in imminently threatened and presently secure areas that will become threatened at uncertain times in the future. The objective was to maximize the expected number of endemic species conserved, predicted with species–area curves. The model was solved for three forms of species–area curve proposed in theoretical studies of habitat loss. Alternative scenarios were considered on the relationship between land prices and development risk. For the most commonly proposed form of the species–area relationship, the fire-fighting approach is optimal even if land prices rise substantially when presently secure areas become threatened. This reflects the assumption that species decline accelerates only after a large proportion of original habitat has been lost. The possibility of large species losses at lower levels of habitat loss justifies at least some pre-emptive conservation, even if land prices are not correlated with threat. If species decline is proportional with habitat loss, the optimal conservation strategy depends strongly on land price dynamics.