One hundred and eighty-one patients attending a dermatology clinic were studied. Twenty-two (12%) were B27 positive. Twenty had peripheral psoriatic arthritis; 3 of these showed sacroiliitis (1 B27 positive, 2 B27 negative). Only one of the other 161 patients had sacroiliitis and he was B27 positive. Subsequently we examined 54 consecutive patients with psoriatic arthritis: 51 had peripheral arthritis (6 with sacroiliitis) and 3 exclusively axial involvement (2 sacroiliitis, one syndesmophytes with normal sacroiliac joints). All patients were pooled together and divided in 4 groups: B27 positive and negative, with or without peripheral arthritis. HLA-B27 and/or peripheral arthritis were associated with an increase in axial involvement. Patients lacking both B27 and peripheral arthritis did not have sacroiliitis and only in one case showed evidence of spinal disease (0.7%). Half of the patients with peripheral arthritis and spinal involvement were B27 positive. All 3 B27 positive patients without peripheral arthritis and with spinal disease were men and they all had bilateral sacroiliitis, indistinguishable from idiopathic ankylosing spondylitis (AS). HLA-B27 and peripheral arthritis appeared to act as separate factors that increased the risk of spinal arthritis in patients with psoriasis. The effect of B27 on psoriasis appeared to be detected in 2 different ways: as a coincidental factor increasing the risk of idiopathic AS (as for the general population) or as one of the multiple HLA associations that increase the risk of psoriatic arthritis; in this latter case the spinal involvement could occur as another manifestation of the clinical course of the disease.