A glass of wine is good for your heart hasn't held up

Don’t drink specifically for heart health. For three decades, a genuine, consistently observed pattern — moderate drinkers appearing to have lower heart disease risk than both heavy drinkers and non-drinkers — supported the idea that a daily glass of wine offered real cardiovascular protection. A better research method designed specifically to rule out the obvious confounding says that protective effect probably was never real.

Why this works

The original finding wasn’t fabricated or sloppy — it showed up repeatedly across large observational studies for over 30 years, producing the well-known “J-shaped curve”: risk highest in heavy drinkers, lowest in moderate drinkers, and somewhat elevated again in non-drinkers. It became one of the most widely repeated pieces of health advice in popular culture, and US dietary guidance for years reflected it directly.

The problem observational studies like this struggle with is confounding — the possibility that moderate drinkers differ from non-drinkers in ways that have nothing to do with alcohol itself but strongly affect heart health. Moderate drinkers, as a group, tend to have higher incomes, better access to healthcare, and generally healthier lifestyles. And critically, “non-drinkers” in many of these studies include former heavy drinkers who quit due to health problems, quietly dragging down that comparison group’s apparent health for reasons unrelated to never having consumed alcohol at all.

Mendelian randomization studies address this by using genetic variants that affect how much people naturally drink — since your genes are assigned essentially at random and can’t be influenced by income, healthcare access, or lifestyle choices, this method sidesteps most of the confounding that plagues standard observational research. Multiple independent Mendelian randomization studies, including a large multi-ancestry analysis from the Million Veteran Program and others conducted separately in European, Chinese, and mixed populations, converge on the same conclusion: the protective J-curve seen in observational data disappears, or in some analyses reverses, when confounding is controlled for this way. A 2024 American Heart Association scientific statement acknowledges this shift directly, and the US Dietary Guidelines Advisory Committee’s own review of multiple Mendelian randomization studies found no support for reduced cardiovascular risk at low drinking levels — a notable, quiet reversal from decades of prior guidance.

It’s also worth being precise about what the newer evidence does and doesn’t say. Nonlinear analyses suggest light drinking is associated with only a small increase in cardiovascular risk, not a large one — this isn’t “light drinking is exactly as dangerous as heavy drinking.” It’s specifically that the claimed protective benefit doesn’t hold up, which is a different and more modest correction than reversing the entire relationship. For cancer risk specifically, the picture is more one-directional: alcohol’s association with several cancers is dose-dependent with no protective dip at low levels in the underlying biology, which was never really in dispute the way the heart disease question was.

How to do it

1. Don’t factor “heart health benefit” into a decision about whether to drink — treat that specific claim as unsupported by the more rigorous current evidence, regardless of what older guidance says.
2. If you enjoy drinking, make that decision based on genuine enjoyment and moderation, not a belief that you’re doing your cardiovascular system a favour by doing so.
3. Keep the actual, better-supported alcohol effects on sleep, mentioned elsewhere on this site, in the calculation instead — those are more directly relevant to how alcohol affects your health day to day.
4. If you currently drink because you believe it’s protective, this is worth genuinely re-examining — not necessarily to stop, but to make the choice for accurate reasons rather than a claim that hasn’t held up.
5. Don’t overcorrect into believing light drinking is now proven equally harmful as heavy drinking — the nuance in the newer evidence is specifically about the absence of benefit, not proof of large harm at low levels.

Common mistakes

• Continuing to cite the J-curve as settled science. It was the dominant view for decades, but the more rigorous causal evidence has moved past it — this is a genuine update, not a fringe opinion.
• Assuming Mendelian randomization studies are some kind of niche methodology. It’s a well-established, widely used technique specifically designed to address the confounding problem that plagues exactly this kind of question.
• Treating “no proven benefit” as “definitely harmful at any amount.” The evidence specifically undermines the protective claim; it’s more nuanced than declaring light drinking equally dangerous as heavy drinking.
• Ignoring that this doesn’t apply to cancer risk, which was never claimed to have a protective dip. The heart disease reversal and the cancer dose-response relationship are two separate findings, not the same story.
• Using this to justify heavier drinking “since moderate isn’t protective anyway.” The heavy-drinking harms were never in question — this reversal is specifically about whether moderate drinking helps, not about whether more drinking is fine.

What should change

Nothing needs to change in terms of a habit to track — the update here is in what you believe alcohol is doing for you, not a new behaviour to monitor. If you’ve been drinking specifically for a perceived heart benefit, that’s the belief worth revisiting.

Your next step

If alcohol is a regular part of your routine, its more directly relevant, better-established effect is on sleep — worth reading that article for a more accurate picture of what a regular drink is actually doing to your body on a night-to-night basis.