Abstract. We present results of a spatially explicit, individual‐based stochastic dispersal model (HexSim) to evaluate effects of size and spacing of patches of habitat of Northern Spotted Owls (NSO; Strix occidentalis caurina) in Pacific Northwest, USA, to help advise recovery planning efforts. We modeled 31 artificial landscape scenarios representing combinations of NSO habitat cluster size (range 4–49 NSO pairs per cluster) and edge‐to‐edge cluster spacing (range 7–101 km), and an all‐habitat landscape. We ran scenarios using empirical estimates of NSO dispersal dynamics and distances and stage class vital rates (representing current population declines) and under adult survival rates adjusted to achieve an initially stationary population. Results suggested that long‐term (100‐yr) habitat occupancy rates are significantly higher with habitat clusters supporting ≥25 NSO pairs and ≤15 km spacing, and with overall landscapes of ≥35–40% habitat. Although habitat provision is key to NSO recovery, no habitat configuration provided for long‐term population persistence when coupled with currently observed vital rates. Results also suggested a key role of floaters (unpaired, nonterritorial, dispersing owls) in recolonizing vacant habitat, and that the floater population segment becomes increasingly depleted with greater population declines. We suggest additional areas of modeling research on this and other threatened species.