Triglyceride to HDL ratio: what it tells you

The ratio uses two numbers from a standard lipid panel, so it costs nothing extra.

In mg/dL (US units): triglycerides ÷ HDL cholesterol. Example: 150 ÷ 50 = 3.0.

In mmol/L (SI units): the same calculation with mmol/L values. Example: 1.69 ÷ 1.29 = 1.31.

The two ratios differ because triglycerides and cholesterol convert with different factors: divide triglycerides in mg/dL by 88.57 and cholesterol by 38.67 (the European consensus rounds these to 88 and 38.6) [1]. So:

• mg/dL ratio × 0.437 = mmol/L ratio
• mmol/L ratio × 2.29 = mg/dL ratio

Always check which units a cut point uses. Even published papers slip: one widely cited study gives 3.0 in mg/dL units and 1.8 as its SI equivalent [3], while direct conversion of 3.0 gives about 1.3.

In overweight adults tested with a direct measure of insulin action, a ratio of 3.0 or more (mg/dL units) identified insulin resistance with 64% sensitivity and 68% specificity, about as well as the full metabolic syndrome criteria. A triglyceride level of 130 mg/dL (1.47 mmol/L)1.47 mmol/L (130 mg/dL) or more did similarly [3].

In a larger follow-up, the ratio was the best routine lipid predictor of both insulin resistance and LDL particle size, with an optimal cut point of 3.5 [4]. Another study found that a ratio of 3.8 separated small, dense LDL (phenotype B) from larger LDL with about 80% accuracy in both men and women [6].

The biology fits: insulin resistance increases VLDL output and apoC-III, which raise triglycerides [7]. A 2024 review of 32 studies with 49,782 people supports the ratio as a simple marker of insulin resistance [8]. For a direct look, see fasting insulin and HbA1c.

No major lipid guideline uses the ratio, so there is no official target. The table lists the main study cut points in both unit systems. The 2024 review found average cut points of 2.53 for women and 2.8 for men [8]. In the Jackson Heart Study, the authors suggested raising the commonly used threshold of 2.0 to 2.5 for African-American men [5].

• In the Jackson Heart Study, the ratio predicted insulin resistance in African-American men (area under the curve 0.77) but too weakly in women (0.66) to define any cut point [5].
• In a UK study, the ratio tracked insulin resistance in White European men and women and South Asian men, but not in South Asian women [2].
• The 2024 review found better accuracy in Caucasian, Asian and Hispanic groups than in African Americans, and in men than in women [8].

The likely reason is that triglycerides run lower in some groups despite insulin resistance; in the Jackson Heart Study, women had a mean triglyceride level of 94 mg/dL [5]. A normal ratio does not rule out insulin resistance, so fasting insulin or a glucose test is a better check when there is doubt.

Some public voices argue that the ratio predicts heart disease better than LDL cholesterol (see perspectives on the hub). The genetic evidence points instead to particle number: coronary benefit follows the change in ApoB whichever lipid moves [9], and in a multivariable analysis only ApoB kept a robust effect [10].

The two can disagree sharply. Among 548 adults on carbohydrate-restricted diets, those with the lowest ratio had the largest LDL rises; a subgroup had triglycerides of 70 mg/dL or less, HDL of 80 or more and LDL of 200 mg/dL or more [11]. A low-carbohydrate diet does improve the ratio on average, lowering triglycerides by 0.26 mmol/L and raising HDL by 0.14 mmol/L compared with low-fat diets [12].

Read the ratio with ApoB and LDL cholesterol, not instead of them.