Why Some Foods Fill You Up and Others Leave You Wanting More

In 1995, a graduate student at the University of Sydney named Susanna Holt did something unusual for a nutrition researcher: she sat people down with isocaloric portions of 38 common foods, asked them to rate their fullness every 15 minutes for two hours, and then offered them as much of a buffet as they cared to eat. She wanted to know, food by food, how long the satiety actually lasted.

The numbers she produced have been quietly circulating in nutrition circles ever since. Boiled potatoes scored 323 on her Satiety Index — more than three times as filling as white bread, the reference point at 100. Fish scored 225. Oatmeal scored 209. At the other end, croissants came in at 47. Cake at 65. Doughnuts at 68. Ounce for ounce and calorie for calorie, some foods quieted hunger for hours. Others left people reaching again before the next meal arrived.

The pattern that emerges from Holt's data, and from the larger body of satiety research it helped launch, is not mysterious. A small set of food properties explains most of the variation. Understanding them is one of the few ways an individual can make calorie counts work less brutally against their own biology.

Protein is the heaviest lever

Across nearly every controlled feeding study, protein produces more satiety per calorie than carbohydrate or fat. Margriet Westerterp-Plantenga at Maastricht University has spent decades cataloguing this difference. Her group's meta-analyses estimate that gram for gram, protein generates roughly twice the satiety response of carbohydrate and substantially more than fat.

The mechanism is multi-channel. Protein triggers the gut to release peptide YY and GLP-1 — two of the satiety hormones — more vigorously than the other macronutrients. It also raises diet-induced thermogenesis: roughly 20–30% of protein calories are spent on digesting and metabolising the protein itself, compared with 5–10% for carbohydrate and 0–3% for fat. And amino acids appear to directly signal fullness through hypothalamic sensing pathways.

The practical implication is one of the most replicable findings in nutrition research. When Heather Leidy at Purdue ran a series of breakfast studies in young adults, switching from a typical high-carbohydrate morning meal (around 13g protein) to a higher-protein version (around 35g) reduced both reported hunger throughout the morning and energy intake at the evening meal. The participants weren't being told to eat less. The biology was doing the work.

Fibre, water, and the physics of a stomach

Holt's two highest-scoring foods — boiled potatoes and fish — share something less obvious than their nutrient profiles. They take up considerable physical space in the stomach for the calories they contain. Volume matters in ways that pure caloric thinking misses.

Barbara Rolls at Penn State has built a 30-year research programme around what she calls "volumetrics": the principle that energy density (calories per gram) is one of the strongest predictors of how much food a person eats at a given meal. Her group has shown repeatedly that when participants are served meals at lower energy densities — more water, more fibre, more air — they spontaneously eat smaller weights of food but consume substantially fewer total calories. The stomach's stretch receptors register volume. The brain reads volume as fullness. Calories per gram is what determines how long that signal lasts before more food is needed.

Fibre extends the effect through two mechanisms. Soluble fibres absorb water and form viscous gels that slow gastric emptying, which prolongs the period the stomach feels distended. Insoluble fibre adds bulk without adding metabolisable calories. Joanne Slavin at the University of Minnesota has reviewed the satiety literature on fibre extensively; the effect sizes are smaller than for protein but real and consistent across dozens of trials.

Why an apple beats apple juice

Few comparisons illustrate the satiety variables more cleanly than whole fruit versus juice. A medium apple delivers roughly 95 calories with about 4 grams of fibre, considerable water, and the chewing time it takes to work through the flesh. Eight ounces of apple juice delivers about 115 calories with essentially no fibre, no chewing, and a glycaemic profile that floods the bloodstream within minutes.

Studies that have served calorie-matched apples and apple juice to participants find that solid fruit produces meaningfully greater satiety and lower subsequent intake. The fruit hasn't gained any nutritional magic. The juice has lost almost everything that signals fullness — volume, fibre, chewing time, and the slower glucose release that comes with eating the matrix intact.

The hidden variable: eating rate

Foods that require chewing also enforce a slower eating rate, and rate of intake turns out to be a surprisingly powerful satiety variable in its own right. Kathleen Melanson at the University of Rhode Island has done some of the cleanest work on this. When the same meal is consumed faster, satiety hormones — particularly CCK and GLP-1 — have less time to rise before the meal ends. People reliably consume more before the brain registers that they have eaten enough.

Croissants, doughnuts, and the rest of the bottom of Holt's index share this property: they are easy to eat quickly. Soft, low in fibre, calorie-dense per bite, and engineered to slide down before the satiety system catches up. A 400-calorie croissant can be finished in three minutes. A 400-calorie bowl of oatmeal takes most people 10–15 minutes to work through. The hormonal feedback loop has time to engage in one case and not the other.

Glycaemic response and the second hunger

What happens an hour or two after a meal matters as much as what happens during it. Refined carbohydrates — white bread, breakfast cereal, juice, most pastries — produce sharp postprandial glucose spikes followed by reactive insulin responses that can drive blood glucose below the pre-meal baseline. The dip arrives at roughly 60–90 minutes for many people, and it brings hunger with it.

A 2021 study by Patrick Wyatt and colleagues at King's College London used continuous glucose monitors in over 1,000 participants and found that the size of the post-meal glucose dip was a stronger predictor of subsequent hunger and energy intake than the size of the initial spike. People who experienced larger dips reported feeling hungry sooner and ate more at their next meal — about 75 extra calories on average over the following 24 hours. The mechanism explains why a sugary breakfast can leave someone hungrier at 10:30am than they were before they ate.

Foods that produce gentler glycaemic curves — those high in protein, fibre, or fat, or those built around intact whole grains — avoid this rebound hunger entirely.

Putting the variables together

The foods at the top of Holt's index tend to share at least three of these properties: high protein density, substantial volume or fibre per calorie, slower eating rate, and modest glycaemic impact. Boiled potatoes are not high in protein, but they score the highest on volume and chewing time, and their resistant starch tempers the glycaemic curve. Fish hits the protein lever hard. Oatmeal combines fibre, water, and slow eating. Eggs at 150 on the index are protein-dominant and slow to consume.

The foods at the bottom — pastries, cake, doughnuts, white bread — fail on essentially every variable: low protein, low fibre, low water, high energy density, rapid intake, sharp glycaemic response. They are, in the most literal mechanical sense, designed not to produce satiety.

This is one of the more useful frames for thinking about food choices without turning eating into accounting. The same calorie count, distributed across different foods, can leave a person comfortably full for four hours or hungry again within ninety minutes. The arithmetic of energy balance still applies, but the biology that determines whether a person can sustain that arithmetic responds to which calories arrived in what package.

Where this stops working

Satiety science can do useful work at the margin for someone whose hunger and reward systems are roughly calibrated. It is considerably less useful for someone whose biology has shifted into the configuration that produces obesity. The leptin signalling can be compromised, the ghrelin response can be elevated, the reward circuitry can override volumetric satiety in ways the original satiety research did not contemplate.

For people in that situation, food choices still matter — protein and fibre remain useful — but they often will not be enough on their own. Leptin resistance and the broader dysregulation of appetite biology are the reason that behavioural interventions alone tend to underperform in clinical obesity treatment. The satiety variables described here describe what works when the system can respond. GLP-1 receptor agonists change what the system is responding with.

Used together, the principles compound. Eating in a way that maximises satiety per calorie while on a medication that restores appetite signalling tends to produce results that neither approach achieves alone.

Key takeaways

• Susanna Holt's 1995 Satiety Index ranked 38 foods against white bread; boiled potatoes scored 323, croissants scored 47 — a sevenfold range at matched calories.
• Protein produces roughly twice the satiety per calorie of carbohydrate, through stronger PYY and GLP-1 release, higher diet-induced thermogenesis, and direct hypothalamic signalling.
• Energy density (calories per gram) is one of the strongest predictors of meal size; Barbara Rolls' volumetrics research shows people eat similar weights of food but very different calorie totals.
• Fibre extends satiety by slowing gastric emptying and adding non-caloric bulk; intact fruit consistently outperforms juice on satiety even at matched calories.
• Eating rate independently affects satiety — faster intake means hormones rise too late to stop overconsumption.
• Post-meal glucose dips predict subsequent hunger; refined carbohydrates that spike and crash drive earlier return of appetite.
• For dysregulated appetite biology, food choices help but often aren't sufficient; GLP-1 medications restore the satiety signalling that volumetric strategies depend on.