There is accumulating evidence for an increased susceptibility to infection in patients with arthritis. We sought to understand the epidemiology of Toxoplasma gondii infection in arthritis patients in eastern China, given the paucity of data on the magnitude of T. gondii infection in these patients.Seroprevalence of T. gondii infection was assessed by enzyme-linked immunosorbent assay using a crude antigen of the parasite in 820 arthritic patients, and an equal number of healthy controls, from Qingdao and Weihai cities, eastern China. Sociodemographic, clinical and lifestyle information on the study participants were also obtained.The prevalence of anti-T. gondii IgG was significantly higher in arthritic patients (18.8%) compared with 12% in healthy controls (P < 0.001). Twelve patients with arthritis had anti-T. gondii IgM antibodies - comparable with 10 control patients (1.5% vs 1.2%). Demographic factors did not significantly influence these seroprevalence frequencies. The highest T. gondii infection seropositivity rate was detected in patients with rheumatoid arthritis (24.8%), followed by reactive arthritis (23.8%), osteoarthritis (19%), infectious arthritis (18.4%) and gouty arthritis (14.8%). Seroprevalence rates of rheumatoid arthritis and reactive arthritis were significantly higher when compared with controls (P < 0.001 and P = 0.002, respectively). A significant association was detected between T. gondii infection and cats being present in the home in arthritic patients (odds ratio [OR], 1.68; 95% confidence interval [CI]: 1.24 - 2.28; P = 0.001).These findings are consistent with and extend previous results, providing further evidence to support a link between contact with cats and an increased risk of T. gondii infection. Our study is also the first to confirm an association between T. gondii infection and arthritis patients in China. Implications for better prevention and control of T. gondii infection in arthritis patients are discussed.This is an epidemiological survey, therefore trial registration was not required.