Studies using resightings of marked birds typically make use of readily-observable tags that are not retained as well as metal legbands. We review methods for estimating survival rate with open capture-recapture / resighting models when tag loss is not negligible. All methods rely on data from double-banding studies, usually carried out as part of the resighting study by application of metal legbands to all birds marked with alternative markers. When tag loss is homogeneous, the methods of Arnason and Mills (1981) and Pollock (1981) can be used. When rates of tag loss depend on time since marking, then a cohort approach can be used and is similar to the methods appropriate for homogeneous tag loss. In addition, Kremers (1987) and Nichols et al. (1992) developed models for the joint analysis of recapture and resighting data in the presence of tag loss. We emphasize the importance of obtaining recapture data in observation-based studies in which tag loss is likely to be a problem. We discuss the allocation of effort to recaptures and resightings for such studies.