Abstract BACKGROUND AND AIMS Atherogenic index of plasma (AIP), a logarithmically transformed ratio of triglyceride (TG)/high-density lipoprotein cholesterol (HDL-C), is a biomarker of dyslipidemia and cardiovascular disease (CVD) risk. However, its applicability in end-stage kidney disease (ESKD) is not clear. We examined the association of AIP with all-cause and CVD mortality in two cohorts of ESKD patients, one from China and one from Sweden. METHOD In 1276 Chinese incident peritoneal dialysis (PD) patients (median age 50 years, 56% males) and 559 Swedish ESKD patients starting on dialysis (n = 392) or undergoing hemodialysis (n = 109) or PD 58 (median age 58 years, 62% males), baseline AIP and other biochemical and metabolic biomarkers were analysed in relation to mortality during follow-up period of up to 5 years. All-cause and CVD mortality risk were analysed with competing-risk regression models with transplantation as competing risk adjusting for age, sex, smoking, diabetes, serum albumin, hemoglobulin and calendar year. RESULTS Chinese patients median (IQR) BMI 21 (19–23) kg/m2 TG 1.4 (1.0–1.8) and HDL-C 1.0 (0.9–1.3) mmol/L compared with Swedish patients median (IQR) BMI 24 (22–28) kg/m2 TG 1.6 (1.2–2.2) and HDL-C 1.2 (1.0–1.6) mmol/L were significantly younger (P < 0.01) and had lower BMI (P < 0.01), TG (P < 0.001), HDL-C (P < 0.001), and Framingham risk score (P < 0.01) while sex distribution was similar. ‘Chinese patients’: After median 3.6 years of follow-up, all-cause mortality rate was 16%, and 91 (46%) of the 199 deaths were caused by CVD. The highest tertile of AIP was associated with higher all-cause sub-hazard ratio sHR (95% CI) 1.52 (1.05); Fig. 1A and CVD mortality sHR 1.82 (1.02–3.27); Fig. 1B risk, after adjusting for confounders. ‘Swedish patients’: After median 2.1 years of follow-up, all-cause mortality rate was 36%, and 89 (44%) of the 204 deaths were caused by CVD. The highest tertile of AIP was associated with lower all-cause sHR 0.66 (0.47–0.94); Fig. 2A and CVD mortality sHR 0.59 (0.35–0.97); Fig. 2B risk after adjusting for confounders. CONCLUSION The highest tertile of AIP was independently associated with higher mortality risk in Chinese ESKD patients while it was associated with lower mortality risk in Swedish ESKD patients suggesting that AIP is context dependent, which limits is usefulness as a risk-assessment tool for predicting clinical outcomes in ESKD patients.