Studies dealing with the individual survival of birds in open populations usually estimate survival according to capture-recapture models like the Cormack-Jolly-Seber (CJS). In fact, these models estimate local apparent survival (ϕ), which is a combination of the probabilities of true survival (S) and site-fidelity (F), i.e. death and emigration are confounded. These S and F parameters can be estimated by using ‘robust’ models (e.g. Barker's model), which use additional resighting and dead reports data. We aim to compare the results (and associated biological implications) obtained by analysing juvenile and adult survival in a Polish urban population of Blackbirds Turdus merula using both the CJS and Barker models. Our CJS models estimated high ϕ values for both juvenile and adult birds (0.48 and 0.62, respectively). The lower scores for juveniles could be interpreted as low juvenile overwintering survival. By fitting Barker models to the same dataset we determined that juvenile site fidelity was lower than that of adults (0.91 and 0.93, respectively), so natal dispersal was slightly greater than breeding dispersal. The high fidelity causes similarity between apparent survival and true survival parameters (S: 0.51 for juveniles, 0.64 for adults). The results are comparable with data from other urban populations. Thus, using robust models certainly allows one to reduce the noise of movements confounding and/or masking survival probabilities, but one can also determine the individual or environmental variables affecting any of them separately.