There is no licensed treatment for non-alcoholic fatty liver disease (NAFLD), a condition that increases risk of chronic liver disease, type 2 diabetes and cardiovascular disease. We tested whether 15-18 months treatment with docosahexaenoic acid (DHA) plus eicosapentaenoic acid (EPA) (Omacor/Lovaza) (4 g/day) decreased liver fat and improved two histologically-validated liver fibrosis biomarker scores (primary outcomes). Patients with NAFLD were randomised in a double blind placebo-controlled trial [DHA+EPA(n=51), placebo(n=52)]. We quantified liver fat percentage (%) by magnetic resonance spectroscopy in three liver zones. We measured liver fibrosis using two validated scores. We tested adherence to the intervention (Omacor group) and contamination (with DHA and EPA) (placebo group) by measuring erythrocyte percentage DHA and EPA enrichment (gas chromatography). We undertook multivariable linear regression to test effects of: a) DHA+EPA treatment (ITT analyses) and b) erythrocyte DHA and EPA enrichment (secondary analysis). Median (IQR) baseline and end of study liver fat% were 21.7 (19.3) and 19.7 (18.0) (placebo), and 23.0 (36.2) and 16.3 (22.0), (DHA+EPA). In the fully adjusted regression model there was a trend towards improvement in liver fat% with DHA+EPA treatment (β=-3.64 (95%CI -8.0,0.8); p=0.1) but there was evidence of contamination in the placebo group and variable adherence to the intervention in the Omacor group. Further regression analysis showed that DHA enrichment was independently associated with a decrease in liver fat% (for each 1% enrichment, β=-1.70 (95%CI -2.9,-0.5); p=0.007). No improvement in the fibrosis scores occurred. Conclusion. Erythrocyte DHA enrichment with DHA+EPA treatment is linearly associated with decreased liver fat%. Substantial decreases in liver fat% can be achieved with high percentage erythrocyte DHA enrichment in NAFLD. (Hepatology 2014;).