Abstract Background Rheumatoid arthritis (RA) is a heterogeneous disease with established poor prognostic factors such as seropositivity, joint damage, and high disease activity at an early, treatment-naïve stage of disease. However, few studies have examined if specific joint locations are correlated with these factors in such a population. This analysis explored the potential correlation of individual swollen and erosive joints with other disease characteristics at baseline and with remission rates in a post-hoc analysis of the Phase III randomized AGREE study. Methods Methotrexate (MTX)-naïve, erosive, RF- and/or ACPA-positive early RA patients (N = 509) were retrospectively evaluated. Baseline joint swelling was analyzed for large and small joints. Baseline erosions were analyzed for wrist, MCP1–5, IP1, PIP2–5 and MTP1–5. Remission rates were assessed after 6 months of treatment with abatacept (ABA) + MTX (N = 256) or MTX (N = 253). The following statistical tests were used: Chi-Square or Fisher’s exact test (categorical variables); Student’s t-test or Wilcoxon rank-sum test (continuous variables); continuity-corrected Chi-square test (efficacy remission endpoints). Results Baseline swelling was most frequent in wrist (91.9%) and MCP2 joint (89.1%), while baseline erosion was most frequent in MTP5 joint (43.5%). Swollen shoulder was significantly correlated (p < 0.0001) with swelling of almost all other large or medium joints. Baseline swelling in the knee, temporomandibular joint (TMJ), wrist and elbow was highly correlated (p < 0.001) with higher tender and swollen joint counts, higher DAS28(CRP) and higher SDAI and CDAI. Baseline swelling was not correlated with erosion per joint, except for MCP2. The largest difference in mean Boolean remission rates at 6 months was in patients with baseline swollen wrist favoring ABA + MTX (14.0% vs 4.4%; p < 0.001). Conclusions Swelling in the large and medium joints (knee, TMJ, elbow and wrist) was highly correlated with severe disease activity while MCP2 swelling seemed to be correlated with joint damage. The correlation of joint locations at an early, treatment-naïve stage with poor prognostic factors, higher disease activity and joint damage, could establish a rapidly progressing anatomical pattern in early RA. Trial registration: ClinicalTrials.gov NCT00122382, registered July 2005.