Current practices of identifying and treating small indolent thyroid cancers constitute an important but in some ways unusual form of overdiagnosis. Overdiagnosis refers to diagnoses that generally harm rather than benefit patients, primarily because the diagnosed condition is not a harmful form of disease. Patients who are overdiagnosed with thyroid cancer are harmed by the psycho‐social impact of a cancer diagnosis, as well as treatment interventions such partial or total thyroidectomy, lifelong thyroid replacement hormone, monitoring, surgical complications and other side effects. These harms seem to outweigh any putative benefit of knowing about a cancer that would not have caused problems if left undiscovered. In addition to harms to patients, thyroid cancer overdiagnosis leads to significant opportunity costs at a societal level, due to costs of diagnosis and treatment. Unlike many other overdiagnosed cancers, accurate risk stratification is possible with thyroid cancer. At the individual patient level, use of this risk information might support informed choice and/or shared decision‐making, as mandated by clinical ethics frameworks. And this approach might, to some extent, help to reduce rates of diagnosis and intervention. In practice, however, it is unlikely to stem the rising incidence and associated harms and costs of overdiagnosed thyroid cancer, especially in situations where health professionals have conflicts of interest. We argue in this article that thyroid cancer overdiagnosis may be usefully understood as a public health problem, and that some public health approaches will be readily justifiable and are more likely to be effective in minimising its harms.