In the present study, we tested our hypothesis on the role of a DQ-DR haplotype in rheumatoid arthritis (RA) predisposition. Using two groups of patients and controls, one from The Netherlands and one from Switzerland, we found that DQA1*0301-homozygous and DQA1*0301//DQA1*0101/04-heterozygous individuals are highly predisposed to RA in both populations, while DQA1*0101/04-homozygous are not. The DQA1*0301-DRB1*0403/06/07 and DQA1*0301-DRB1*0901 haplotypes are not associated with RA by themselves but strongly increase the risk of developing disease in DQA1*0301- and DQA1*0101/04-heterozygous. DRB1 alleles carrying the motif DERAA in their third hypervariable region, i.e., *0103, *0402, *1102, *1103, *1301, and *1302, provide a long-lasting protection against RA in DQA1*0101/04- but not in DQA1*0301-positive individuals. These data show that considering both DQ and DR gives a better distinction between patients and controls than the shared epitope hypothesis.