The China Study: A Real Dataset, and a Claim Bigger Than the Evidence It Rests On

The China Study was written by T. Colin Campbell, a biochemist and longtime nutrition researcher at Cornell University, together with his son Thomas M. Campbell II, a physician. BenBella Books published it in 2005, and it went on to sell more than a million copies, becoming one of the most widely read books connecting diet to cancer and other chronic disease.

Its foundation is a real, large research project: the China-Cornell-Oxford Project, an observational survey begun in the 1980s that collected dietary, lifestyle and disease-rate data across a number of counties in rural China, with Campbell as one of its lead investigators. That dataset genuinely exists, was genuinely large in scope, and has been used in legitimate peer-reviewed nutrition research beyond this book.

The book's central argument builds on that dataset together with Campbell's earlier laboratory research on casein, the main protein in cow's milk, and its effect on tumor growth in animal studies. The combined case the book makes is broad: that animal protein generally, and casein specifically, promotes the growth of cancer and other chronic diseases, and that a whole-food, plant-based diet can prevent, slow or in some cases reverse them.

This is the title this page's own sourcing guidance names specifically as one where a book's claims run ahead of the evidence, and that assessment deserves an honest, specific explanation rather than a blanket dismissal, because the underlying data is real even where the conclusions drawn from it are contested.

Campbell's academic career included decades of federally funded nutrition research before this book, and he has continued publishing and speaking on the same thesis since, including a follow-up book with his son. That sustained body of work is worth acknowledging: this is not a single opportunistic title but the culmination of one researcher's long-held view, which makes the gap between his confidence and the wider field's more measured position all the more worth examining carefully.

The book opens with Campbell's own research background and the origins of the China-Cornell-Oxford Project, then moves through chapters connecting diet to specific diseases: heart disease, several cancers, diabetes, autoimmune disease, and others, each built around correlations found in the China data alongside supporting laboratory and other research.

A significant portion of the book is devoted to the casein research specifically, describing animal studies in which diets higher in casein were associated with faster tumor growth and diets lower in casein with slower growth, research Campbell conducted earlier in his career and treats as mechanistic support for the broader dietary argument.

Later chapters translate the argument into practical terms: recommendations to minimize or eliminate animal products, minimize processed food and refined carbohydrates, and build meals around whole plant foods, along with guidance on supplementation, since the book acknowledges that a diet excluding all animal products requires attention to nutrients such as vitamin B12.

The book also includes appendices addressing common objections and questions, including concerns about protein adequacy on a plant-based diet, which Campbell addresses directly rather than leaving unaddressed, a structural choice that reflects an awareness of likely reader skepticism even where the underlying causal argument remains contested.

The book is strongest, and its lasting influence rests on, popularizing a genuinely evidence-supported general direction: reducing processed food and refined carbohydrates and building meals around vegetables, fruit, whole grains and legumes overlaps substantially with WCRF's own evidence-graded cancer prevention recommendations. Whatever the dispute over the book's strongest claims about animal protein specifically, its broader push toward whole, minimally processed foods sits within the mainstream of dietary cancer-risk guidance.

The China-Cornell-Oxford Project itself is also a real contribution to nutrition science. Large-scale observational surveys of this kind are a legitimate and important part of how researchers generate hypotheses about diet and disease, even when, as with any observational data, they cannot on their own establish that a specific dietary factor causes a specific disease outcome.

This is where the book's claims run ahead of what its own evidence can support, and it is worth being precise rather than sweeping about why.

The China-Cornell-Oxford Project is an ecological, observational study: it compares disease and dietary patterns across regions and populations rather than tracking individuals over time under controlled conditions. That design can reveal genuine correlations worth investigating further, but it cannot, on its own, establish that one specific dietary factor causes a specific disease, because regions differing in diet also differ in many other ways, genetics, infection rates, environmental exposures, economic conditions, that can independently affect disease rates and confound the association.

Independent reviewers outside Campbell's own research group have raised this concern directly. Critics have argued that the book's references do not fully support the strength of the claims made from them, and that the book does not adequately explain data within its own dataset that runs against its thesis, including notably elevated stomach cancer rates in some of the surveyed counties that do not track cleanly with the book's animal-protein hypothesis. That critique does not mean the China-Cornell-Oxford data is worthless or fabricated; it means the book's causal conclusions reach further than an observational, ecological dataset can reliably support on its own.

The casein research, while real laboratory science, was conducted primarily in animal models under controlled diets far more extreme in composition than typical human eating patterns, and translating dose-dependent effects in a rat's diet directly into an argument against animal protein in human diets generally is a bigger leap than the book's confident framing suggests.

NCI's own material on diet and cancer risk reflects this same broader caution about observational nutrition research: it notes that human population studies of diet and cancer generally show associations rather than proof of cause and effect, and that even where laboratory research on a specific compound looks promising, human evidence has often failed to confirm the same effect. That is precisely the gap between the book's laboratory-and-observational evidence and the strength of its central causal claim.

This book is one of the clearest cases available for practicing the distinction between association and causation, because its underlying data is genuinely substantial and its central causal leap is genuinely larger than that data can bear alone.

A stronger cancer claim, in NCI's framing, typically rests on evidence from randomized controlled trials or, where trials are not feasible, on multiple independent, well-designed observational studies whose findings converge and that account carefully for confounding factors, with the distinction between association and causation stated explicitly rather than blurred. A weaker claim tends to rest on a single dataset, a compelling personal or laboratory narrative, and language that moves quickly from correlation to cause.

WCRF's Cancer Prevention Recommendations are a useful, more conservative benchmark precisely because they are built by an independent scientific panel reviewing the accumulated global evidence and updated as new research arrives, rather than resting on one team's dataset and interpretation. WCRF supports limiting red and processed meat specifically, as one part of a broader dietary pattern, a narrower and more precisely evidenced claim than the book's wider case against animal protein as a category. WCRF does not recommend eliminating animal protein generally, and its position on dairy specifically is more measured than the book's casein-focused argument.

None of this is a reason to dismiss the book's broader nutritional direction, which points toward genuinely supported territory: more vegetables, fruit, whole grains and legumes, less processed food. The caution is narrower and more specific: the book's strongest, most attention-getting claim, that animal protein and casein specifically are a primary driver of cancer, rests on evidence, real evidence, that is weaker than the confidence with which the book presents it. The appeal of that stronger claim is understandable; a single, identifiable dietary villain is a simpler and more actionable story than the messier, multi-factorial picture the wider evidence actually supports, and that is worth naming honestly rather than mocking.

For the evidence-graded picture directly, see diet and cancer risk, and for a related look at another single-factor diet claim treated the same way, see can keto or carnivore diets prevent cancer.

It suits a reader interested in the history and scope of large-scale nutrition epidemiology, and one willing to read its central causal claim alongside independent critique rather than at face value. It suits someone looking for a push toward a more plant-forward, whole-food diet, understood as one reasonable direction among several rather than as the single proven answer to cancer risk.

It does not suit a reader looking for a settled scientific consensus on animal protein and cancer, which the book's confident framing implies more strongly than the underlying evidence, including data from its own study, actually supports. Anyone considering eliminating an entire food category from their diet, particularly during active cancer treatment, is better served discussing that change with a registered dietitian on their care team first.

Cancer Explained is free and ad-free, and you can support this work if it helped.

• NCI — Diet
• World Cancer Research Fund — Cancer Prevention Recommendations
• NCI — Complementary and Alternative Medicine

This page discusses The China Study for education. It is not medical advice, and nothing here is a judgement of anyone's real medical care.