Suggestive, but largely unproven, observations have associated arthritis with environmental antigens, including foods. We studied a patient with inflammatory arthritis in a prospective, "blinded," controlled fashion to determine whether her symptoms were associated with food sensitivities. This 52-year-old white woman with 11 years of class I, stage I, active disease, had symptomatic exacerbations allegedly associated with meat, milk, and beans. We observed an increase in symptoms following an unblinded food challenge and then studied her in our clinical research unit. On her normal diet for 6 days, she averaged 30 minutes of morning stiffness, 9 tender joints, 3 swollen joints, 87% subjective assessment (100% = best possible), and 89% examiner assessment. While she was fasting (3 days) or taking Vivonex (2 days), we noted no morning stiffness, tender joint score of 1, swollen joint score of 0, and assessments of 100% (P less than 0.05 versus normal diet). She was then nourished with Vivonex for 33 days without difficulty and challenged in a blinded fashion at mealtimes with lyophilized foods placed into opaque capsules. Four milk challenges (equivalent to greater than or equal to 8 ounces per meal) produced up to 30 minutes of morning stiffness, 14 tender joints, 4 swollen joints, subjective assessment of 85%, and objective assessment of 80% (P less than 0.05 versus fasting-Vivonex), peaking 24-48 hours postchallenge. Placebo and other foods (lettuce and carrots) were without effect.(ABSTRACT TRUNCATED AT 250 WORDS)