Good Calories, Bad Calories — Gary Taubes (2007)

One-line verdict: A forensic dismantling of the diet-heart hypothesis and the low-fat dietary consensus, arguing that the last 50 years of nutrition science and public health policy were built on methodologically weak foundations — and that carbohydrates, not fat, drive obesity, heart disease, and metabolic illness.

Who should read this: Anyone who wants to understand why the official nutrition consensus looks the way it does, and whether it deserves the authority it's been granted. It rewards readers willing to work through detailed epidemiological and biochemical history; it is not a diet book, despite what the title implies.

The Central Argument

Taubes's thesis has two interlocking parts. First, the dietary-fat/heart-disease hypothesis — the idea that saturated fat raises cholesterol and cholesterol causes heart disease — was accepted not because the evidence was strong, but because influential advocates (primarily Ancel Keys) suppressed contradictory data and steamrolled critics. Second, and more consequentially, the real drivers of obesity, type 2 diabetes, heart disease, and several cancers are refined carbohydrates and sugar, operating through the mechanism of insulin. Fat doesn't make you fat; insulin does, by directing cells to store rather than burn energy.

The strongest evidence for the first claim is historical: the Seven Countries Study, which Keys used to establish the fat-heart link, cherry-picked nations and excluded data from countries that didn't fit. The strongest evidence for the second claim is biochemical: insulin's role in regulating fat storage and release (lipogenesis vs. lipolysis) is not contested — the argument is that public health ignored these mechanisms in favor of a simpler calorie-in/calorie-out model.

Key Ideas

The carbohydrate-insulin hypothesis of obesity. Fat accumulation is regulated by insulin, which is driven primarily by carbohydrate consumption. Eat refined carbs → insulin spikes → fat cells store energy and resist releasing it → hunger increases even as fat accumulates. This is a hormonal model, not a behavioral one. Taubes insists obesity isn't caused by gluttony or sloth; those are symptoms, not causes.

The calorie hypothesis is unfalsifiable and probably wrong. "Eat less, move more" follows logically only if you assume all calories are metabolically equivalent. But the macronutrient composition of food affects hormones, satiety, and energy expenditure differently. Taubes argues the calorie model was adopted for its simplicity, not its evidentiary basis.

Ancel Keys and the politics of nutrition science. Keys was a forceful, politically adept scientist who identified fat as the dietary villain as early as the 1950s and spent decades ensuring the scientific establishment agreed. He attacked opponents personally (most notably John Yudkin, who argued sugar was the culprit), controlled key advisory committees, and shaped the dietary guidelines that emerged from the McGovern Senate Select Committee in 1977. Taubes documents this in detail — this is the book's most damning and original section.

The McGovern Report as a turning point. The 1977 dietary guidelines recommending reduced fat and increased carbohydrate consumption were issued before the supporting science existed. They were driven partly by agricultural economics, partly by political urgency, and partly by premature scientific consensus. What followed was arguably a natural experiment: Americans complied broadly with low-fat advice and got sicker.

The quality of nutrition epidemiology is genuinely poor. Diet-disease studies are almost always observational, rely on self-reported food-frequency questionnaires, and are confounded by the "healthy user bias" (people who follow dietary advice tend to have healthier behaviors generally). The effect sizes being measured are typically within the margin of error. Taubes spends considerable time making this methodological case rigorously.

Insulin resistance as the common thread. Most of the chronic diseases Taubes covers — type 2 diabetes, hypertension, certain cancers, Alzheimer's disease — share insulin resistance as a mechanism or correlate. The hypothesis is that a carbohydrate-heavy diet produces chronic hyperinsulinemia, which in turn produces insulin resistance, which then manifests in multiple disease pathways.

Traditional populations and the natural experiment of westernization. Numerous indigenous populations — Inuit, Masai, Plains tribes — ate near-zero-carbohydrate diets with high fat and protein and had essentially no obesity, diabetes, or heart disease until they adopted Western diets. Their decline tracked the introduction of refined carbohydrates, not fat. Taubes treats these as observational evidence, not proof, but the pattern is striking.

The Framingham Heart Study's unreported findings. Framingham is the bedrock study used to establish the cholesterol-heart disease link. Taubes documents that the study's own director, William Castelli, acknowledged that among men over 50 (the largest portion of the cohort), higher cholesterol was associated with lower cardiovascular mortality — a result that was not prominently published and did not make it into the public health messaging derived from the study.

LDL is not a monolith. Not all LDL particles behave the same. Small, dense LDL is highly atherogenic; large, buoyant LDL is largely benign. The dietary intervention that raises LDL most reliably (eating saturated fat) tends to raise the large, benign fraction. The dietary intervention that raises small, dense LDL is high carbohydrate consumption. This distinction was largely absent from public health messaging.

Frameworks & Vocabulary

Lipogenesis / Lipolysis: The biochemical processes of fat storage and fat burning, respectively. Both are primarily regulated by insulin. This framing repositions obesity as a disorder of fat partitioning rather than caloric surplus.

Hyperinsulinemia: Chronically elevated insulin levels, produced by persistent carbohydrate consumption, especially refined carbohydrates and sugar. Taubes treats this as the proximate cause of insulin resistance and metabolic syndrome.

The diet-heart hypothesis: The claim that dietary fat (especially saturated fat) raises serum cholesterol, which causes atherosclerosis and heart disease. Taubes's main target — he argues it was never adequately demonstrated and has driven policy catastrophically.

The carbohydrate hypothesis of obesity and disease: Taubes's proposed alternative. Refined carbohydrates and sugar are the dietary villains; the mechanism is insulin; the diseases are metabolic in origin.

Metabolic syndrome: The cluster of conditions (abdominal obesity, hypertension, high triglycerides, low HDL, insulin resistance) that tend to co-occur and share a likely common cause. Taubes argues this syndrome, not simple high cholesterol, is the real predictor of cardiovascular risk.

Strongest Evidence & Stories

The Seven Countries Study. Keys collected data from 22 countries but reported only the 7 that supported his hypothesis. When all 22 are included, the correlation between saturated fat and heart disease largely disappears. This is not a minor methodological quibble — it is the foundational study of the entire dietary fat consensus.

John Yudkin's suppression. Yudkin, a British physiologist, published Pure, White and Deadly in 1972 arguing that sugar, not fat, was the primary dietary driver of heart disease. Keys attacked him viciously and effectively. The scientific community largely abandoned the sugar hypothesis for two decades. Taubes's account of this episode reads as a case study in how scientific consensus can be manufactured by dominant personalities rather than converging evidence.

The Women's Health Initiative (2006). The largest randomized controlled trial of the low-fat diet — nearly 50,000 women followed for eight years — found that the low-fat diet produced no reduction in cardiovascular disease, no reduction in cancer, and essentially no weight loss compared to controls. It was published the year before Taubes's book and is his most powerful piece of contemporary evidence. The mainstream nutritional establishment's response to its null findings was, in Taubes's telling, to explain them away rather than question the underlying hypothesis.

Tensions, Limitations & What the Author Gets Wrong

The book is 600 pages, and probably 400 would have done it. The middle chapters on cancer and dementia are thinner than the cardiovascular and obesity sections. The case that carbohydrate drives these diseases is circumstantial and Taubes presents it with less rigor than the central argument deserves.

Taubes is a journalist, not a scientist, and it shows in one specific way. He is excellent at prosecuting a case against bad science but does not fully grapple with the evidence against his own hypothesis. There are randomized controlled trials showing benefit from reducing saturated fat in specific high-risk populations. There are populations (e.g., traditional Asian societies) with very high carbohydrate consumption and very low rates of metabolic disease — a fact he handles briefly and not entirely satisfactorily.

The carbohydrate-insulin model has since been tested more rigorously — and the results are mixed. The NuSI-funded studies (partly motivated by Taubes's work) did not consistently show the metabolic advantages of very low-carbohydrate diets that the hypothesis predicts. Calories appear to matter more than Taubes suggests, even if carbohydrate quality matters too. His dismissal of the energy-balance model is probably too complete.

The book was written before the "food reward" and ultra-processed food hypotheses gained traction. Hyperpalatable, engineered food products may drive overconsumption through mechanisms (palatability, reward circuitry) that are orthogonal to the carbohydrate-insulin pathway. Taubes's model doesn't account for this well.

The narrative structure favors the prosecution. Taubes is skilled at finding the most damaging version of evidence against the dietary-fat consensus. He applies less of this skepticism to his own preferred hypothesis. A reader should approach this as a corrective brief, not a balanced account.

How This Connects

The most direct intellectual predecessor is John Yudkin's Pure, White and Deadly (1972), which Taubes rehabilitates. Robert Lustig's subsequent work on fructose and sugar is essentially a biochemical elaboration of Taubes's argument. Peter Attia, who later became one of the most influential voices in longevity medicine, credits this book directly as formative.

The methodological critique of nutrition epidemiology anticipates John Ioannidis's broader work on research reliability and the replication crisis. Thomas Kuhn's Structure of Scientific Revolutions is the obvious framing for Taubes's account of how paradigms resist disconfirmation — though Taubes doesn't invoke it explicitly.

The Uncomfortable Implication

If Taubes is even partially right, the 1977 dietary guidelines were one of the most consequential policy errors in modern public health history — not because the science was uncertain (uncertainty is unavoidable), but because the guidelines were issued with false confidence, enforced by institutional authority, and insulated from falsification. The obesity epidemic and the explosion of type 2 diabetes that followed may be at least partly iatrogenic: caused by the very advice meant to prevent them.

The harder implication is institutional. The same structures that produced that error — expert consensus, government dietary guidelines, industry funding of nutrition research, the suppression of heterodox scientists — remain intact. Taubes isn't just making a claim about fat and carbohydrates. He's making a claim about the conditions under which bad science becomes policy and stays policy long after the evidence should have dislodged it.