We studied the dispersal behavior of 1,475 northern spotted owls (Strix occidentalis caurina) during banding and radio-telemetry studies in Oregon and Washington in 1985-1996. The sample included 324 radio-marked juveniles and 1,151 banded individuals (711 juveniles, 440 non-juveniles) that were recaptured or resighted after dispersing from the initial banding location. Juveniles typically left the nest during the last week in May and the first two weeks in June (x?? ?? SE = 8 June ?? 0.53 days, n = 320, range = 15 May-1 July), and spent an average of 103.7 days in the natal territory after leaving the nest (SE = 0.986 days, n = 137, range = 76-147 days). The estimated mean date that juveniles began to disperse was 19 September in Oregon (95% CI = 17-21 September) and 30 September in Washington (95% CI = 25 September-4 October). Mean dispersal dates did not differ between males and females or among years. Siblings dispersed independently. Dispersal was typically initiated with a series of rapid movements away from the natal site during the first few days or weeks of dispersal. Thereafter, most juveniles settled into temporary home ranges in late October or November and remained there for several months. In February-April there was a second pulse of dispersal activity, with many owls moving considerable distances before settling again in their second summer. Subsequent dispersal patterns were highly variable, with some individuals settling permanently in their second summer and others occupying a series of temporary home ranges before eventually settling on territories when they were 2-5 years old. Final dispersal distances ranged from 0.6-111.2 km for banded juveniles and 1.8-103.5 km for radio-marked juveniles. The distribution of dispersal distances was strongly skewed towards shorter distances, with only 8.7% of individuals dispersing more than 50 km. Median natal dispersal distances were 14.6 km for banded males, 13.5 km for radio-marked males, 24.5 km for banded females, and 22.9 km for radio-marked females. On average, banded males and females settled within 4.2 and 7.0 territory widths of their natal sites, respectively. Maximum and final dispersal distances were largely independent of the number of days that juveniles were tracked. Although statistical tests of dispersal direction based on all owls indicated that direction of natal dispersal was non-random, the mean angular deviations and 95% CI's associated with the samples were large, and r-values (vector length) were small. This lead us to conclude that significant test results were the result of large sample size and were not biologically meaningful. Our samples were not large enough to test whether dispersal direction from individual territories was random. In the sample of radio-marked owls, 22% of males and 44% of females were paired at 1 year of age, but only 1.5% of males and 1.6% of females were actually breeding at 1 year of age. At 2 years of age, 68% of males and 77% of females were paired, but only 5.4% of males and 2.6% of females were breeding. In contrast to the radio-marked owls, most juveniles that were banded and relocated at 1 or 2 years of age were paired, although few were breeding. Although recruitment into the territorial population typically occurred when owls were 1-5 years old, 9% of banded juveniles were not recaptured until they were > 5 years old. We suspect that our estimates of age at recruitment of banded owls are biased high because of the likelihood that some individuals were not recaptured in the first year that they entered the territorial population. A minimum of 6% of the banded, non-juvenile owls on our demographic study areas changed territories each year (breeding dispersal). The likelihood of breeding dispersal was higher for females, young owls, owls that did not have a mate in the previous year, and owls that lost their mate from the previous year through death or divorce. Mean and median distances dispersed by adults were