Twenty-six patients were operated because of osteo-arthritis of the trapeziometacarpal joint. After excision of the trapezium, they were randomized to receive either a Swanson silicone trapezium implant or a tendon interposition arthroplasty using a strip of the abductor pollicis longus tendon (APL). No infection or clinical silicone synovitis occurred and no major radiographic cyst formation was found. Two endoprostheses dislocated early. All 13 patients in the Swanson group and 11 of the 13 in the APL group were subjectively satisfied at 2-5 year (mean, 43 months) follow-up. All patients were free of pain at rest and at light work, but half of the patients in each group still experienced pain with heavy work. Thumb pinch strength and range of motion were not significantly different in the two groups. The trapezial space decreased both with load and length of follow-up in both groups. Five endoprostheses subluxed during stressed pinch.In conclusion, both methods gave good, but not complete, pain relief and neither produced better results than the other in the short term.