We estimated annual return rate, fidelity, and breeding dispersal in a migratory population of Flammulated Owls (Otus flammeolus) in central Colorado. Return rates, based on capture-recapture histories of 39 males and 52 females from 1981 to 2003, were higher for males (84%) than for females (45%). Annual recapture probability was higher for females, because breeders are easier to capture than nonbreeders and females always attempted to nest, whereas some males were unpaired (did not nest) for up to four years. Territory fidelity was male biased (92%, vs. 56% for females, adjusted for undetected emigration), and mean tenure on territories was more than twice as long for males as for females. Females, but not males, had lower return rates to territories in the year following nesting failure compared with females whose nests were successful. Most males appeared to occupy one territory their entire reproductive lives, countering predictions of habitat-selection models that individuals should move to higher-quality habitats when they become available. We estimated that 74% of pairs retained the same mate in consecutive nesting attempts, but mates that bred together for multiple years had no reproductive advantages over mates that bred together for the first time. In most cases, females dispersed from territories if their mates did not return. When females dispersed, they went to territories where total productivity over the study and lifetime reproductive success of new mates were higher than on original territories, which supports the hypothesis that dispersal by females increases individual fitness.