Cognitive impairment was found related to higher mortality among the elders. However, study results about cognitive malfunction at earlier stages of specialist assessment or treatment to mortality for the ones without dementia were still in debate. An anonymised electronic database in South London and Maudsley NHS Foundation Trust (SLAM) covering 1.2 million population was utilised to assess the influence of cognitive impairment measured by Mini-Mental State Examination (MMSE) to survival among the clients 65+ years old when firstly assessed during 2007-2010. They were followed up for survival till the end of July, 2011 by the linkage to National Health Services in UK. Overall and subgroup analyses for specific diagnoses were done by Cox regressions. Age, gender, psychiatric diagnosis, ethnicity, marital status, primary care trust, and deprivation score were considered as potential confounders. A total of 7,196 subjects were identified, including 3,674 subjects diagnosed as dementia, 1,435 as depression and 492 as delirium. The adjusted hazard ratios of MMSE score in quintiles (30- 28, 37-25, 24-21, 20-16 and 15-0) when firstly assessed in SLAM were 1.19, 1.36, 1.69 and 2.13 (95% CIs: 0.99-1.42, 1.15- 1.62, 1.42-2.01 and 1.79-2.55) with 1st quintile as reference group (p-value of test for trend < 0.01). Except the delirium group, analyses for all the ones but dementia excluded, the depression group, and depression group with dementia excluded revealed similar outcomes. In current study, we identified a universal effect of cognitive impairment to mortality. Declining cognition function might reflect underlying physical conditions leading to death.