It is commonly understood that the cumulative impact of increased habitat destruction on biological systems does not scale linearly. Despite this, current environmental decision-making strategies often fail to incorporate these non-linearities because impact assessments are regularly performed independently without considering contributions to future cumulative impacts, which are, instead, externalised to the environment. Here we used a stochastic modelling framework to examine the effect of three decision-making strategies - externalised, individually-borne and shared cumulative costs - on the number of species driven to extinction by the development of a region. We also tested how different levels of vulnerability (quantified as the correlation between development benefits and species richness) and irreplaceability (expressed as the average area of occupancy of an assemblage) influence the outcomes of development. Overall, the inclusion of cumulative impacts in our simulations resulted in the destruction of fewer patches and, as a consequence, more modest benefits obtained from developments. Moreover, these patterns were strongest when cumulative costs were shared by all developers. The most striking finding was that the distinction between decision-making strategies was highest in landscapes of high vulnerability and irreplaceability. Environmental decision-makers are recommended to commit to a strategy for incorporating the cumulative costs of development because, as this study suggests, this affects the endpoints of conservation and development considerably; especially in the most vulnerable and irreplaceable landscapes.