An ultrasonic tagging program for tuna was conducted in 1988 and 1989 within the Regional TunaProject of the Indian Ocean Commission. Three yellowfin and six skipjack tuna were tagged withtemperature or depth sensitive transmitters in the North-western part of the Mozambique Channel(12°S-44°E) around Anjouan island (Comoros Archipelago) where several fish aggregating devices(FADs) were previously moored. The horizontal and vertical movements observed during 8 tracks (3yellowfin and 5 skipjack tuna) whose duration was between 3 and 24 hours, are analysed in terms ofswimming depth, temperature encountered and position of the tracked tuna relative to the FAD orcoast line. Comparison between recorded depth of tracked tuna and echo sounded fish indicatedtracked tuna were schooling.Two of the 3 tagged yellowfin tuna displayed a behaviour of association with FADs. The optimaldistance between 2 anchored FADs, to avoid adverse interference in the attraction of tuna, is estimatedas 11 nautical miles. A very small percentage of time is spent by yellowfin tuna near the surface. Themean swimming depths encountered in the daytime by yellowfin tuna are much deeper (70-110 m)than they are at night (40-70 m). The relative homogeneity in the observed behaviour of yellowfintuna and the fair general agreement with previous results obtained in the Pacific Ocean, should allowapplication of ultrasonic tagging results to fishing and prospecting purposes in the future.The movements of the 5 tracked skipjack tuna do not indicate a behavioural association withFADs, and do not present marked differences between the swimming depths encountered by nightand during the daytime. The high variability observed in the behaviour of the different trackedskipjack tuna, and the bad agreement with previous results obtained in the Atlantic and in the PacificOceans have to be emphasized. A high turnover of the skipjack tuna concentrated around FADs dueto an intense and pelmanent migratory flow through the area of Comoros Islands could partly explainthese apparent discrepancies.