Cognitive impairment in the elderly is a key issue affecting quality of life. Despite of a strong genetic background, a limited number of SNPs have been found in the Western population which explains only small proportions of the genetic components. The cognitive ability of an individual is reflected by different domains each representing a specific aspect of cognition like perception, memory, reasoning, and attention. Current genetic studies on cognitive abilities are based on association with each individual domain (test) separately or with a total score which is a simple sum up of all test scores on different domains. Multivariate association analysis that takes into account pleiotropic genetic effects jointly on the different domains with increased statistical power is actually missing. This project aims at (1) applying multivariate classical twin models to estimate genetic correlation among different domains of cognitive function using Chinese twins; (2) introducing and performing the first multivariate GWAS on cognitive performances in old age Chinese twins, in search of genetic variants that are pleiotropic for different domains of cognition and domain-specific genetic variants. Results from this exploratory study will provide candidate SNPs of cognitive aging, for future validation in the Chinese population, and for comparison across populations.