Background: Epidemiological studies and randomised clinical trials (RCTs) report disparate findings in relation to omega-3 polyunsaturated fatty acids (n-3 PUFA) benefit for cardiac patients. With RCTs interpretation is potentially confounded by background n-3 PUFA intake. The goal of this pilot, small cohort, pre-surgical supplementation study was to evaluate post-operative atrial fibrillation (AF) and cardiac molecular expression profiles employing two data analysis approaches - by treatment randomisation and by stratification using measured n-3 PUFA. Methods: Patients (n=20) received 3g/day of fish or placebo oil (FO vs PO) in a double blind randomised protocol prior to elective coronary artery graft and valve surgery. Groups were matched for age, gender, and mean treatment duration (~20 days). Resected atrial myocardium was sampled for assay of viability metabolic markers, and blood obtained for erythrocyte membrane lipid measurement. Results: There was substantial overlap of cell membrane n-3 PUFA content across PO and FO groups, and no group treatment effects on AF incidence or myocardial molecular marker levels were detected. In contrast, data stratification using membrane n-3 PUFA content (at 8% total membrane lipid) achieved significant separation of patients (by n-6:n-3 PUFA ratio), a significant differential cardiac expression of the marker peroxisomal proliferator-activated receptor, but no difference in AF incidence. Conclusions: This small n-3 PUFA case study demonstrates that the same cohort may yield differing findings when evaluated using randomisation or stratification approaches based on direct molecular measures in cell membranes.