The Sugar Dose-Response

Everyone agrees that a lot of added sugar is bad. The harder question — how much is too much — has a messier answer than the headlines suggest, because the evidence is shaped like a ramp, not a cliff. Here is what the famous mortality curve actually shows, what each daily soda adds up to, and how the official thresholds were set and honestly read.

Why the Dose Question Is Harder Than It Looks

In an ideal world there would be a trial that randomized people to 10, 25, and 50 grams of added sugar a day for decades and counted deaths. That trial will never run — nobody consents to it, and nobody funds it. What exists instead is a two-layer stack: prospective cohort studies that measure sugar intake and wait, and short randomized trials that measure the intermediate machinery — weight, triglycerides, liver fat, insulin sensitivity. The cohorts carry the hard endpoints but are observational; the trials are causal but short and soft. The honest dose-response read therefore requires both, plus a hard look at confounding, because high sugar intake travels with low diet quality, smoking history, and sedentary behavior — all of which statisticians can adjust for imperfectly. That is why every number on this page is an association with adjustment, and none of them is a threshold with a cliff.

The Famous Mortality Curve

The study most quoted in the "how much is too much" debate linked added-sugar intake in a nationally representative US sample to mortality records, following roughly 31,000 adults for a median of about 15 years (Yang et al., JAMA Internal Medicine, 2014). Added sugar was expressed as a share of daily calories, and the result was a staircase: compared with people under 10%, those at 10–24.9% had a 30% higher risk of cardiovascular death, and those at 25% or more had a 175% higher risk — hazard ratios of 1.30 and 2.75 after adjustment for the usual suspects. About one in ten US adults sat in that top band at the time. Two honest glosses belong right next to those numbers: the top category starts at an intake most people would call extreme, and the confidence interval on the 2.75 is wide — the point estimate is dramatic, the precision is not.

What Each Serving of Soda Adds

Below the top band, the cleaner signal comes from beverages. A meta-analysis of 17 cohorts found that each additional daily serving of sugar-sweetened beverage was associated with 18% higher type 2 diabetes risk before adjusting for body weight, and 13% after (Imamura et al., BMJ, 2015). The gap between those two numbers is itself informative: a large share of the harm runs through weight gain, and a share doesn't. Earlier meta-analyses told the same story for the metabolic syndrome (Malik et al., Diabetes Care, 2010). On the trial side, the causal slice is about calories: systematic reviews of randomized trials find that cutting sugary drinks produces modest, consistent weight loss, mediated almost entirely by reduced energy intake (Te Morenga et al., BMJ, 2013). Do the serving math once and the arithmetic explains itself: a standard 330ml can delivers roughly 35–40 grams of sugar — about 140–160 calories with zero satiety attached, which is why it adds to the day rather than replacing part of it.

The Thresholds on the Books

The famous ceilings are budgets, not pharmacological thresholds. They were set by expert committees reasoning backward from calorie math and cohort risk — no experiment located a cliff at any of them.

Note the family resemblance: all three are percentage rules wearing gram clothes, and all three sit far below what the average adult in a Western country actually consumes. The practical reading is that the consensus ceiling is genuinely low — a single sweetened coffee drink can spend the day's entire budget.

Reading the Epidemiology Honestly

• 📈 The curve is a ramp, not a cliff. Risk climbs through the middle band; the data cannot localize a threshold, so any "X grams is safe" claim is invented precision.
• 🔄 Substitution sets the context. What replaces the sugar matters — the fats and carbs topic covers the swap arithmetic, and swapping soda for water beats swapping it for juice.
• 🥤 Liquids carry most of the evidence. Beverages are where the cleanest, most consistent signals live — solid-food sugar at equal doses is less studied.
• 🔍 Confounding is real but survivable. The dose-response gradients persist after adjustment, which is the main reason the field takes them seriously.
• 🔢 Absolute risk stays modest. Tripling a small risk still leaves a small risk — the top band's hazard ratio is about the worst-case end of the curve, not the typical end.

A Workable Personal Line

• 🥤 Beverages at zero, or near zero. The highest-value single rule in the whole literature — the parent topic ranks it first for the same reason.
• 🧾 Ten grams per serving. A packaged snack above ~10g added sugar per serving is dessert, whatever the front of the box says.
• 🍰 Keep the treats, shrink the dose. Small, savored, after a meal — dessert as occasion, not default.
• 🚶 Blunt what you do eat. A short walk after meals measurably shaves the post-meal glucose bump.
• 🌾 Partner sugar with fiber and protein. The fiber and protein topics explain why the same grams inside a meal behave differently.

Two Days, Worked

Take one person and two weekdays. Thursday: a sweetened latte on the way in (about 30 grams), a granola bar mid-morning (12), a soda with lunch (40), pasta with a jarred sauce at dinner (6 more, hiding in the sauce). The day closes somewhere near 90 grams of added sugar, and nothing about it felt like a binge — it felt like a normal busy day. Saturday: black coffee, yogurt with berries, a tuna sandwich, a stir-fry whose sauce contributes 4 grams, and one square of dark chocolate after dinner — under 20 grams total, including the treat. Same person, same week, a fivefold gap — and the gap is not virtue, it's defaults. The latte route, the vending machine, the jarred sauce: Thursday was engineered, Saturday wasn't. Which is exactly why the working line is built on changing the defaults, not on being a person who never wants sugar.

Questions, Answered Briefly

• 🎯 Is 25 grams a hard line? No — it's a budget anchored to about 5% of calories. Risk climbs gradually, and the data can't localize a cliff anywhere on the ramp.
• 🥤 What about one daily soda? That's roughly 40 grams — it spends the entire day's budget in one drink. The serving-by-serving meta-analyses make this the clearest single avoid on the page.
• 🍎 Do natural sugars count? Intrinsic sugars in milk and whole fruit don't carry the added-sugar signal; the guidelines target free and added sugars specifically.
• 🧃 Does 100% juice count? In the WHO's definition, yes — juice concentrates free sugars into liquid form. Treat it as a sugar-sweetened beverage with vitamins.
• 📊 Should I track grams daily? Only if curiosity helps you. The signal worth tracking is the big sources — beverages and packaged snacks — not the last gram of sauce.

Related Topics