Objective To systematically review the evidence of the factors associated with self-rated health (SRH) and estimating the population attributable risk proportion (PARP) of interests in the elderly. Methods MEDLINE, EMbase, CBM, CNKI, VIP and WanFang Data were searched (from January 1960 to April 2011) for relevant literature. The combined effect on each factor associated with poor SRH was expressed as RR or OR (95%CI). The contribution of each factor to poor SRH in the elderly population was estimated with PARP. Meta-analysis was performed using RevMan 5.1 software. Results Twelve studies involving a total of 35 349 participants aged 60 or above were eligible in this systematic review. Ten studies were conducted in China and the rest two studies were from the USA and Brazil. Of them, only one was prospective cohort study and 11 studies were cross-sectional. The results of meta-analysis showed that the main factors associated with poor SRH in the elderly were dependency, diabetes and coronary heart disease with the corresponding OR and 95%CI as 12.10 (6.31, 23.20), 6.49 (3.21, 13.09) and 5.60 (1.07, 29.42), respectively. However, the top three factors contributing to poor SRH in the elderly population were coronary heart disease, having illness half one year ago and chronic disorders with the corresponding PARP as 53.91%, 52.56%, and 50.09%, respectively. It was not possible to perform sub-group analysis because of the limited quantity of studies on each factor associated with poor SRH. Conclusion Based on the current available evidence, it appears that chronic disorders are closely related to poor SRH and contribute most in the elderly population. However, due to the limitation of the included predominantly cross sectional studies which are not enough for causal argument, it is insufficient to determine the association of many factors with poor SRH. Further high-quality prospective cohort studies are needed.