Forty-nine patients with disseminated gonococcal infection (DGI) hospitalized at Boston City and University Hospitals over a 7-year period were studied. Patients with clinical manifestations of DGI and with cervical, urethral, rectal, pharyngeal, synovial or blood cultures positive for Neisseria gonorrhoeae were separated into two groups based on the presence or absence of suppurative arthritis. There were 19 cases of suppurative arthritis (Group II) and 30 cases with only tenosynovitis, skin lesions, or both (Group I). Blood cultures were positive only in Group I patients (43%) and synovial fluid cultures only in Group II patients (47%). Polyarthralgia was the most common initial symptom in both groups of patients. Twenty-six Group I patients had tenosynovitis (87%), while only 4 Group II patients (21%) had tenosynovitis (p less than 0.001). The knee was the most commonly involved suppurated joint. Twenty-seven Group I patients (90%) had skin lesions compared to 8 Group II patients (42%) (p less than 0.001). Some of these lesions progressed on treatment; some patients were unaware of their lesions. Genitourinary symptoms were unusual in both groups of patients. Eleven women (33%) were menstruating or were pregnant at the onset of DGI. Thirteen patients had histories suggestive of previous gonococcal infections; one had recurrent DGI. This patient and one other were found to have complement abnormalities. There were no cases of endocarditis or meningitis. Four patients had unexplained liver function abnormalities. All patients recovered uneventfully. Strains isolated from disseminated sites were predominantly of the transparent phenotype (90%). Many strains (58%) required arginine, hypoxanthine and uracil for growth. They were also more susceptible to penicillin than reported strains that cause pelvic inflammatory disease. Most strains were of a single outer membrane protein coagglutination serogroup, WI (85%). These characteristics did not vary between the Group I and Group II isolates. The two groups of strains, however, did vary in their complement-dependent bactericidal reactivity to normal human sera. Eighteen of 24 Group I strains (75%) versus 9 of 19 Group II strains (47%) resisted killing by all normal human sera tested (p less than .05). Likewise, convalescent sera from Group II patients were able to kill their infecting strains more often than did sera from Group I patients (70% vs 17%) (p less than 0.01). Thus, variations in the clinical expression of disease in patients with DGI may be explained, in part, by differences in certain phenotypic and immunologic features of infecting strains.