The loss of external tags is one of the biggest problems in mark-recapture research. An evaluation of tag loss is therefore required to improve tagging methodology, select appropriate tag types, and accurately estimate population status and dynamics. We estimated tag loss probability of double-tagged green turtles (Chelonia mydas) nesting at Redang Island, Malaysia (05°49′ N 103°00′ E), from 1993–2014. For both titanium and Inconel tags, we found a tag loss pattern with a high initial tag loss rate that decreased to an asymptote near a constant value above zero. The initial tag loss probability for titanium tags was higher than for Inconel tags, and titanium tags were lost earlier than Inconel tags in more individuals when both types of tags were attached. In addition, comparison of the tag loss probability of Inconel tags attached during the period when tagging staff changed every year to that when senior tagging staff were not changed, indicated that lack of consistency in tagging staff affected the application of at least one of two Inconel tags. Estimated tag loss probability was incorporated into a recapture model, and annual survival probability was estimated to be 0.893 (95% confidence interval: 0.857–0.921).