All natural populations are confronted with the temporal variability of their environment, and most of them occur in fragmented habitats. I use spatially explicit modeling to examine the interactive effects of habitat configuration, habitat deterioration and spatially correlated environmental perturbations on the viability of fragmented populations. When considering a fixed amount of habitat, viability is maximized for an intermediate (optimal) density of habitat patches, allowing effective dispersal without strong environmental correlation among patches. Starting from this optimal density, I show that the scale of habitat change (density reduction vs range contraction) interacts with the direction of change (reduction vs improvement of habitat availability). An improvement in habitat availability is always more beneficial if occurring through an increase of the species range, while a reduction in habitat availability is always more critical if occurring through a reduction of patch density (even in the presence of environmental correlation). In the context of the ‘single large or several small’ debate in reserve design, results indicate that a large number of small patches may be optimal to long-term species persistence if the species range increases with the number of patches.