OBJECTIVE To determine normal limits for erythrocyte sedimentation rate (ESR) in rheumatology clinics based on observations from patients with noninflammatory disorders (NID); to determine the proportion of patients with osteoarthritis (OA) excluded from clinical trials because of elevated ESR; to determine the proportion of patients with rheumatoid arthritis (RA) meeting ESR criteria for remission, clinical activity, and eligibility for clinical trials; and finally, to explain elevations of ESR in OA. METHODS Cross sectional and longitudinal study of all rheumatic disease clinic outpatients with RA (N = 1,556, ESR = 12,683) and NID (N = 3,961, ESR = 5,706). RESULTS For all NID the 90th, 95th percentiles were 33, 40 for women and 23, 31 for men. For patients with OA, 21.2% of women and 8.5% of men had ESR > or = 30 mm/h. ESR in women with OA but not men with OA or those with RA were significantly associated with body mass index. Twenty-nine (29.4) percent of men and 41.6% of women with RA satisfied the ESR remission criterion. When the active disease criterion was considered (ESR > or = 28 mm/h), only 54.5% of men and 62.6% of women, on the average, have active RA. CONCLUSIONS The upper limit for normal ESR for women through age 60 is about 38 mm/h. A significant proportion of patients with RA with active disease will satisfy the ACR ESR criterion for remission, but only 54-63% of patients being treated in a rheumatology clinic will have active disease (ESR > or = 28 mm/h). Current use of the ESR as a criterion in clinical trials and remission criteria while based on wide clinical experience is contradictory and may not reflect actual data.