Objective. To compare 3 sets of criteria for meaningful improvement in a rheumatoid arthritis (RA) clinical trial, and to evaluate the implications of these criteria sets for RA trial design. Methods. Data were obtained from the Minocycline in Rheumatoid Arthritis (MIRA) trial (primary outcome measures: 50% improvement in joint tenderness and 50% improvement in joint swelling, based on joint scores). These MIRA data were evaluated against 1) the Paulus criteria (20% improvement in 4 of 6 measures: joint tenderness scores, joint swelling scores, physician's and patient's global assessments, erythrocyte sedimentation rate [ESR], and morning stiffness); and 2) the American College of Rheumatology (ACR) criteria (20% improvement in joint tenderness and joint swelling counts, and in 3 of 5 other measures: physician's and patient's global assessments, ESR, modified Health Assessment Questionnaire, and patient's pain assessment). The ACR criteria were modified using 3 of 4 remaining measures, since baseline pain assessment data were not available. Results. Percentages of minocycline-treated patients versus placebo-treated patients showing meaningful improvement were as follows: by MIRA criteria, for joint tenderness, 56% versus 41% (P = 0.021), and for joint swelling, 54% versus 39% (P = 0.023); by Paulus criteria, 41% versus 28% (P = 0.040); and by ACR criteria, 44% versus 26% (P = 0.004). Both the modified ACR criteria and the Paulus criteria demonstrated a reduced placebo response rate. Compared with the MIRA criteria, the ACR criteria increased, and the Paulus criteria decreased, absolute between-group differences in improvement; however, both criteria sets increased relative percentages of patients showing improvement in the minocycline group versus the placebo group. Study design considerations indicated that application of the ACR criteria would reduce the required sample size. Conclusion. Different placebo response rates and treatment group differences were found using the 3 RA improvement criteria sets. These findings support the use of the ACR criteria for defining improvement in RA clinical trials.