Only around 180 breeding pairs of the endangered Short-tailed Albatross Diomedea albatrus remain. The overwhelming majority of these, about 170 pairs, nest on Tori-shima, in the Izu Islands (30°28'N, 140°18'E), an uninhabited island about 600 kilometers south of Tokyo. The remaining pairs nest in the Senkaku Islands. Tsubame-zaki, the traditional breeding site on Tori-shima, is located on a steep slope composed of loose volcanic gravel. This poor environment is thought to be responsible for the low breeding success rates observed in recent years. The project described here was designed to improve breeding success by relocating the colony to Hatsune-zaki on the other side of the island, a more favorable breeding site with gentler slopes.A preliminary experiment was conducted near the traditional colony at Tsubame-zaki in November 1991, using 10 elaborate life-like decoys and tape recorded albatross calls. Result showed that the Short-tailed Albatrosses are attracted by these decoys and calls. A second experiment was conducted for four days in April 1992 at Hatsune-zaki, using 16 decoys and tape recorded calls. Nine individuals were observed to fly close over the study site. In November 1992, 41 decoys were set up at Hatsune-zaki for the 1992-93 breeding season, with no vocal lure. In March 1993, 9 decoys were added, and a long-term playback device, driven by solar batteries, was installed as a vocal lure. The number of decoys was increased steadily, reaching a total of 69 in November 1994. Every year from 1993, observations at the study site were conducted during daylight hours for a period of about six days either in January, February or March. Five individuals were observed to land in March 1993, and 5, 29, and 41 individuals were observed to land in 1994, 1995, and 1996 respectively. The total time of stay per unit observation time (stay index) increased from 0.31 in 1993 to 0.86 in 1994, 1.64 in 1995, and 2.37 in 1996. Stay index thus increased 7.6 times during this four year period. Percentage of landings in the area of decoys to all the landings at Hatsune-zaki was 78.9% in 1993, but declined to 23.1% in 1994, 8.3% in 1995, and 31.4% in 1996, with more landings occurring in the area below the decoys. It was suggested that changes in the vocal lure apparatus and directions of speakers made in 1993 were responsible for this switch in landing area. When landing in the area of decoys, birds more often landed near, or approached on foot, adult-type decoys than sub-adult types. Also, landings tended to concentrate in the area where the volume of play-back sounds was high. Most of the birds landing at the site were immature and sub-adult. Only four adults out of a total 41 landings were observed in 1996. Four to six year old birds accounted for 70% of all the landings in 1995. Courtship displays by two individuals were observed in 1993, and in 1994 an individual was observed to stay overnight at the study site. Courtship displays were again observed several times in 1994; and two individuals were observed to remain at a fixed spot for a long time in 1995. In November 1995 a six year old male and a five year old female bred at the site, producing a single egg which was hatched successfully. This young left the island on 10 June 1996. In November 1996, three pairs built nests, two of which produced eggs. One of the two successful pairs was the pair that bred in 1995. Three of the four breeding birds were of known age, and had been born in 1990 or 1991. Ages at first breeding for these three birds were 4 years 11 months (female), 5 years 11 months (male), and 6 years 11 months (male). Individuals that were hatched at Tsubame-zaki in 1990 were first observed at Hatsune-zaki in January 1994. These observations suggest that it takes two to three years for an individual to start breeding after the first arrival at the breeding site.