Post-operative cognitive dysfunction is defined as a decline in cognitive functions that occurs after surgery, but different diagnostic criteria and incidences have been reported in medical literature. Our aim was to determine incidence of post-operative cognitive dysfunction 3 months after non-cardiac surgery in adults. A systematic review of available evidence was performed by PRISMA guidelines. A search was done in May-July 2015 on PubMed, EMBASE, CINAHL, LILACS, Scielo, Clinical Trials, and Grey Literature Reports. Inclusion criteria were prospective design studies with patients over 18 years old, surgery under general or regional anesthesia, follow-up for 3 months, and use of a neurocognitive battery for diagnosis. We excluded studies made on cardiac or brain surgery patients. Risk of bias was assessed using tools from National Heart Lung and Blood Institute. We selected 24 studies. Average age was 68 years. Only five studies reported incidence of cognitive decline for a non-surgical control group. Median number of tests used was 5 (range 3-13). Pooled incidence of post-operative cognitive dysfunction at 3 months was 11.7% [95% CI 10.9-12.5] but with several methodological differences between studies. Increasing age was the most consistent risk factor identified (seven studies). Post-operative cognitive dysfunction in patients is frequent, especially in patients over 60 years old. Limitations include methodological differences in studies. Efforts must be made to reach a consensus in definition and diagnosis for future research.