Composition of dietary fatty acid intake, which influences cytokine production, may contribute to the development of non-Hodgkin’s lymphoma (NHL). Serum lipid levels may serve as biomarkers of inflammation associated with NHL risk. We conducted a case–control analysis (275 cases and 549 controls) nested within the Multiethnic Cohort Study (whites, Japanese Americans, Latinos, African Americans, and Native Hawaiians) to examine the association of prediagnostic, erythrocyte membrane phospholipid fatty acid composition, and serum cholesterol and triglyceride (TG) concentrations with the risk of NHL. Conditional logistic regression was used to calculate odds ratios (OR) and 95 % confidence intervals (CI) by tertiles of biomarker concentrations. Higher total saturated fatty acids (SFA) were associated with an increase in NHL risk (ORT3 vs. T1 = 1.57 [95 % CI: 1.03–2.39]; p trend = 0.01), whereas no associations were detected for total n−3 or n−6 polyunsaturated fatty acids. Inverse associations were observed for total cholesterol (TC; OR T3 vs. T1 = 0.51 [95 % CI: 0.35–0.74]; p trend < 0.0001) and high-density lipoprotein cholesterol (HDL-C; OR T3 vs. T1 = 0.47 [95 % CI: 0.31–0.71]; p trend = 0.0001) but not for low-density lipoprotein cholesterol or TG. Adjustment for the use of lipid-lowering medication did not modify the results substantially. This prospective biomarker investigation offers supportive evidence for an adverse effect of higher erythrocyte membrane SFA levels on NHL risk, but preclinical effects cannot be excluded. Inverse relations between prediagnostic, circulating TC and HDL-C and NHL risk may be due to reverse causation or a result of protective actions of these lipids and lipoproteins.