Comparison of fish size by different sampling methods (commercial and training vessels and scientific observer) for Japanese longline fishery operating in the Indian Ocean was conducted to examine representative- ness of size data and to consider how to apply to stock assessment models. Size data by training vessels, which operated mainly between 1960s and 1980s in the tropical area of eastern Indian Ocean, were main component during this period. Size data measured by scientific observers have been main component since mid-2000s espe- cially for bigeye tuna. Length frequencies of the fish in the same area-quarter strata were usually similar among sampling methods if sufficient number of fish were measured, although some differences were observed. In sever- al strata a mode of smaller fish was observed only as for the fish measured by training vessels and/or scientific observers. Difference of average weight of the fish between based on catch and effort data and size data was ob- served by about 10 kg or more for a part of period. Some considerations and examinations will be necessary to decide how to apply size data to stock assessment models.