Hypogammaglobulinaemic patients admitted to the Medical Research Council's immunoglobulin therapy trials up to 1963 were examined for evidence of arthritis and were compared with population samples from Leigh, Wensleydale and Watford which were used as controls. 2. Four out of 60 males had inflammatory polyarthritis compared with an expected 0.4. In addition, 3 had synovitis limited to the knees and 4 had had polyarthritis in the past, the expected figures being 0.12 and 2.4 respectively. None of the patients had definite rheumatoid arthritis as defined by the American Rheumatism Association Criteria, though one satisfied the New York Criteria for Still's disease. All were seronegative and none of the patients with polyarthritis had radiological evidence of arthritis. None of the 18 female patients had arthritis. The prevalence of arthritis of synovitis in males was 29% in 1964, but in the following 3 years varied between 8 and 11%, suggesting a strong environmental influence. 3. All of the 7 males with inflammatory polyarthritis had levels of IgG, IgA, IgM and IgD, which were less than 30% of the standard normals and steatorrhoea was present in three. Steatorrhoea was also noted in 2 of 5 patients wih arthritis who died before the start of the survey, compared with 7% of all the males. 4. Clinical inflammatory polyarthritis was no more common in the first-degree relatives of patients with hypogammaglobulinaemia than in the population as a whole and they had no excess of erosive arthritis visible in X-rays of the hands or feet nor was sacro-iliitis excessively frequent. 5. Positive tests for rheumatoid factor were found in only 2 of the patients with hypogammaglobulinaemia and the first-degree relatives had no more positive sheep cell agglutination or latex fixation tests than expected in random population samples. 6. It is concluded that the seronegative polyarthritis which occurs in patients with hypogammaglobulinaemia is unrelated genetically to rheumatoid arthritis or Still's disease.