A wide variety of attachment techniques have been used to track birds with electronic tags, with glue, tape, leg rings, neck collars and harnesses being the most common methods. In general, the choice of attachment method should strive to minimize tagging effects, but ensure that sufficient data are collected to address the research question at hand. The aim of our study was to develop and evaluate tag attachment methods to track Sandwich Terns Thalasseus sandvicensis during the last part of the incubation and the chick-rearing period of one breeding season. Tag attachments had to stay on for the duration of the chick-rearing period (5–6 weeks) and be non-restraining and flexible, but strong enough to withstand the forces and submersion associated with their plunge-diving foraging technique. We first experimentally tested the durability of flexible material under various environmental conditions with the aim of developing a self-releasing harness. Then, in field studies, we compared three different attachment methods on terns during the breeding seasons, attaching tags to dorsal feathers using (1) tape, (2) glue or (3) a newly developed harness made specifically for short-term deployments of one chick-rearing period and constructed from degradable material. Assessment of the performance of attachment methods was based on retention time of the loggers and on annual survival rates of tagged individuals in comparison with non-tagged individuals. The use of tape and glue led to premature loss of tags (median minimum retention time (range) of 3 (1–4) days and 15 (5–26) days, respectively), whereas the self-releasing harness had a median minimum retention time of 42 (18–91) days, which is sufficient to track Sandwich Terns during the entire chick-rearing period. The apparent annual survival of birds tagged using glue or tape did not differ from that observed in non-tagged control birds. In contrast, birds fitted with the self-releasing harnesses might have experienced a lower survival rate than control birds. Entanglement of birds in the harness material was incidentally observed in three cases, which may have contributed to the lower survival rates observed in this group. The risk of entanglement can potentially be mitigated with a leg-loop harness instead of a full-body harness. Our results highlight the necessity of careful consideration when selecting appropriate attachment methods. Specifically, there is a need to address whether the research questions and desired tracking duration justify the use of a harness and the higher impact that it entails, or whether a tape or glue-mount is sufficient. More broadly, sharing field expertise in tag attachments across studies is essential to successful deployments while minimizing the impact on animals.