Estimating bycatch is essential for monitoring the ecological impacts of a fishery in order to set management and mitigation priorities. Purse seine vessels targeting tropical tunas incidentally catch pelagic sharks (mainly silky and oceanic whitetip sharks), which are brought onboard and can be observed on the upper and lower decks. Currently, single onboard observers can only be efficiently stationed on one of the two decks, and thus often rely on information provided by the crew to complement their bycatch estimations. In this study, we used dedicated scientists strategically positioned during fishing sets in order to establish a reference count of captured sharks during conventional commercial fishing trips. We then assessed the accuracy of the counts made by (i) single observers onboard during the same fishing trips in the Pacific Ocean (where observers' main duty is to estimate catch of target species and bycatch estimation is of a lower priority) and the Atlantic Ocean (where observers' focus is on bycatch) and (ii) Electronic Monitoring System (EMS) in the Indian Ocean. A total of 74 fishing sets conducted during four purse seine fishing trips revealed that shark counts were underestimated for 50%-100% of the sets, with the mean shark count underestimation, at the fishing trip level, ranging from 9% to 40% (onboard observers) and 65% for EMS. Given the importance of monitoring populations of vulnerable species, we strongly encourage specific studies during which the complementary counts of two onboard observers are used simultaneously to assess the accuracy of various EMS configurations, bearing in mind that single onboard observers appear to underestimate the number of captured sharks.