The traditional methods of identifying biomarkers in rheumatoid arthritis (RA) have focussed on the differentially expressed pathways or individual pathways, which however, neglect the interactions between pathways. To better understand thepathogenesis of RA, we aimed to identify dysregulated pathway sets using a pathway interaction network (PIN), which considered interactions among pathways. Firstly, RA-related gene expression profile data, protein–protein interactions (PPI) data and pathway data were taken up from the corresponding databases. Secondly, principal component analysis method was used to calculate the pathway activity of each of the pathway, and then a seed pathway was identified using data gleaned from the pathway activity. A PIN was then constructed based on the gene expression profile, pathway data, and PPI information. Finally, the dysregulated pathways were extracted from the PIN based on the seed pathway using the method of support vector machines and an area under the curve (AUC) index. The PIN comprised of a total of 854 pathways and 1064 pathway interactions. The greatest change in the activity scorebetween RA and control samples was observed in the pathway of epigenetic regulation of gene expression, which was extracted and regarded as the seed pathway. Starting with this seed pathway, one maximum pathway set containing 10 dysregulated pathways was extracted from the PIN, having an AUC of 0.8249, and the result indicated that this pathway set could distinguish RA from thecontrols. These 10 dysregulated pathways might be potential biomarkers for RA diagnosis and treatment in the future.