Habitat destruction and fragmentation increasingly bring humans into close proximity with wildlife, particularly in urban contexts. Animals respond to humans using nuanced anti-predator responses, especially escape, with responses influenced by behavioral and life history traits, the nature of the risk, and aspects of the surrounding environment. Although many studies examine associations between broad-scale habitat characteristics (i.e., habitat type) and escape response, few investigate the influence of fine-scale aspects of the local habitat within which escape occurs. We test the “habitat connectivity hypothesis,” suggesting that given the higher cost of escape within less connected habitats (due to the lack of protective cover), woodland birds should delay escape (tolerate more risk) than when in more connected habitat. We analyze flight-initiation distances (FIDs) of five species of woodland birds in urban Melbourne, south-eastern Australia. A negative effect of habitat connectivity (the proportion of the escape route with shrubs/trees/perchable infrastructure) on distance fled was evident for all study species, suggesting a higher cost of escape associated with lower connectivity. FID did not vary with connectivity at the location at which escape was initiated (four species), apart from a positive effect of habitat connectivity on FID for Noisy Miner Manorina melanocephala. We provide some support for two predictions of the “habitat connectivity hypothesis” in at least some taxa, and conclude it warrants further investigation across a broader range of taxa inhabiting contrasting landscapes. Increasing habitat connectivity within urban landscapes may reduce escape stress experienced by urban birds.