In an attempt to develop a safe, proliferating, oral, attenuated vaccine against shigellosis, genes that control the synthesis of group- and type-specific somatic antigens of Shigella flexneri 2a were transferred via conjugation to a recipient strain of Escherichia coli. The resultant hybrid (E. coli expressing shigella surface antigens) vaccine strain, PGAI 42-1-15, believed to have a complete (smooth) lipopolysaccharide, was given to volunteers in two vaccination-challenge studies. The vaccine was well tolerated and gave evidence of intestinal proliferation. In trial no. 1, volunteers given two doses of vaccine one month apart were challenged after eight weeks with 10(4) virulent S. flexneri 2a. Attack rates were comparable in vaccinees (50%) and controls (40%). In trial no. 2, vaccinees were given three weekly doses of vaccine and were challenged four weeks later with a small inoculum (10(2)) of S. flexneri 2a. Again, attack rates among vaccinees (47%) and controls (39%) were similar. It is unclear why this theoretically ideal, live shigella vaccine failed to protect against S. flexneri 2a.