Characteristics of Canadian RA patients started on anti-tumor necrosis factor (TNF) treatment were compared with 12 other countries. Data from the Optimization of HUMIRA trial (OH) were compared with Canadian real world studies [Ontario Biologics Research Initiative (OBRI) and the Real-Life Evaluation of Rheumatoid Arthritis in Canadians Receiving HUMIRA (REACH)], and to data from American, Australian, British, Czech, Danish, Dutch, Finnish, German, Italian, Norwegian, Spanish, and Swedish RA databases. Patient characteristics and temporal trends at initiation of anti-TNF therapy were compared between countries. Baseline Disease Activity Scores (DAS28) varied from 5.3 to 6.6. Lower disease severity was noted in databases from countries with less restrictive anti-TNF coverage: Dutch [based on previous disease-modifying antirheumatic drugs (DMARD) use, DAS28, swollen joint count (SJC), tender joint count (TJC), Health Assessment Questionnaire Disability Index (HAQ-DI), Danish (previous DMARD use, DAS28), Norwegian (DAS28, SJC, TJC, visual analog scale (VAS) of global health), and Swedish (DAS28, SJC, TJC, HAQ-DI)]. RA databases showed lower disease scores than did OH (P < 0.05). The US databases also showed lower disease severity (CORRONA: previous DMARD use, SJC, TJC; National Data Bank for Rheumatic Diseases: HAQ, P < 0.001). The UK and Czech Republic had restrictive coverage and higher mean baseline DAS28 than OH (P < 0.001). Baseline DAS28 in the registries with published data lowered over time (British, Norwegian, Danish, and Swedish) but less for the British (P < 0.001). These results confirm that regional variation exists between the 13 countries analyzed in the initiation of treatment with anti-TNF agents among RA patients and suggest that in some cases this variation may be increasing. In some countries the mean baseline disease severity declined over time and regional reimbursement policies and differences in physician preferences may be influencing initiation of anti-TNF therapy in RA.