Between February 1993 and July 1997, 150 horses suffering from recurrent uveitis were subjected to parsplana vitrectomy. In these horses, antibody titers to Leptospira serovars were determined in serum samples and in samples from diluted vitreous collected during vitrectomy. Although the vitreous samples were diluted with 250 ml of balanced salt solution, in 86 of the 150 vitreous samples (= 57%) the antibody titers were higher than in the serum samples. Additionally, serum samples from 77 horses suffering from ERU, but which were not subjected to vitrectomy, and serum samples from 97 horses with clinically normal eyes were analyzed for antibodies to Leptospira serovars. Among the 227 horses with ERU (150 treated surgically, 77 treated conservatively) 50 horses (50 of 227 = 22%) had serum antibody titers to Leptospira serovars of > or = 1:800. Among the 97 horses with clinically normal eyes, 24 horses (24 of 97 = 25%) had serum antibody titers to Leptospira serovars of > or = 1:800. In undiluted vitreous samples from 20 horses with clinically normal eyes, no antibody titers to Leptospira serovars could be detected. Among the 150 horses with ERU, 90 animals (90 of 150 = 60%) had antibody titers of > or = 1:100 in the diluted vitreous samples, the difference being highly significant (p < 0.001). The findings are discussed in relation to the etiology of recurrent uveitis in horses.