With growing and ageing populations, the incidence of dementia is expected to triple globally by 2050. In the absence of effective drugs to treat or reverse the syndrome, dietary approaches which prevent or delay disease onset have considerable population health potential. Prospective epidemiological studies and mechanistic insight from experimental models strongly support a positive effect of a high fish and long chain n-3 fatty acid (EPA and DHA) intake on a range of cognitive outcomes and dementia risk, with effect sizes equivalent to several years of ageing between the highest and lowest consumers. As reviewed here, an effect of EPA and DHA on neuroinflammation and oxylipin production is likely to in part mediate the neurophysiological benefits. However, randomised controlled trials (RCTs) with EPA and DHA supplementation have produced mixed findings. Insight into the likely modulators of response to intervention and factors which should be considered for future RCTs are given. Furthermore, the impact of APOE genotype on disease risk and response to EPA and DHA supplementation is summarised. The prevalence of dementia is several-fold higher in APOE4 females (about 13% Caucasian populations) relative to the general population, who are emerging as a subgroup who may particularly benefit from DHA intervention, prior to the development of significant pathology.