Objective Psoriatic arthritis (PsA) is a common inflammatory joint disease distinct from other chronic arthritides and frequently accompanied by psoriasis vulgaris. In a first genome-wide association study (GWAS), we were able to identify several genetic risk factors. However, even combined with previously identified factors, the genetic contribution to disease was not fully explained. Therefore, we undertook this study to investigate further 17 loci from our GWAS that did not reach genome-wide significance levels of association in the initial analysis. Methods Twenty-one of 22 single-nucleotide polymorphisms were successfully genotyped in independent cohorts of 1,398 PsA patients and 6,389 controls and in a group of 964 German patients with psoriasis vulgaris. Results Association with a RUNX3 variant, rs4649038, was replicated in independent patients and controls and resulted in a combined P value of 1.40 x 108 by Cochran-Mantel-Haenszel test and an odds ratio (OR) of 1.24 (95% confidence interval [95% CI] 1.151.33). Further analyses based on linkage disequilibrium (LD) at RUNX3 refined the most significant association to an LD block located in the first intron of one isoform. Weaker evidence for association was detected in German patients with psoriasis vulgaris (P = 5.89 x 102; OR 1.13 [95% CI 1.001.28]), indicating a role in the skin manifestations of psoriasis. Conclusion Our analyses identified variants in RUNX3 as susceptibility factors for PsA. RUNX3 has already been implicated in susceptibility to ankylosing spondylitis, another spondyloarthritis, although its risk allele is independent from the one for PsA. RUNX-3 is involved in CD8+ T lymphocyte differentiation and is therefore a good candidate for involvement in PsA and psoriasis vulgaris as T cellmediated diseases.