Fifty radio-tagged long-tailed bats Chalinolobus tuberculatus, belonging to three social groups numbering on average 72–132 bats each, ranged over an area of 11 700 ha in Nothofagus rainforest in the Eglinton Valley, Fiordland, New Zealand, during the austral summers of 1993–96. Bats were followed for 12.7 ± 6.3 (sd) days. Home-range sizes were among the largest published for Microchiroptera. Median minimum convex polygons (100% MCPs) for adult males were 1589 ha (max = 5629 ha), 1361 ha for post-lactating females, and 657 ha for non-reproductive females. Ranges of lactating females (median = 330 ha), were significantly smaller. Juveniles that were volant for < 2 weeks had the smallest ranges (median = 237 ha) but movements increased significantly after c. 2 weeks to a median of 2006 ha. Movements were frequent and rapid within the range (790 ± 471 sd m/15 min) and average range lengths were 3.3–10.9 km (max = 19 km). Despite their large home ranges, bats concentrated their activity (85% of fixes) in small core areas that represented a mean of 5.7 ± 1.5% (sd) of the 100% MCPs. Ranges of roosting sites were also small, averaging 9.4 ± 1.8% (sd) of total range size. Bats followed similar movement patterns each night but core areas only overlapped by 43 ± 14% (sd) from one night to the next. Moderately low levels of overlap of home ranges (31–68%) and of core areas (23–56%) among individuals of the same reproductive class suggest a degree of spatial segregation among bats from the same roosting group. A prediction that C. tuberculatus has large ranges to minimize overlap between foraging bats, reflecting scarcity of food, requires testing. Range sizes were probably underestimates of seasonal, annual, and life-time range requirements. Nevertheless, large range size and the degree of individual spacing implies that conservation areas designed for bats should be large and that protection of roosting areas alone would not be sufficient to protect bat populations.