Most proposed definitions of recurrent implantation failure (RIF) are based on clinical judgement, probably affected by patients’ demands. They are not based on robust statistical considerations. As a result, a diagnosis of RIF is commonly made too early, exposing couples to the risk of overdiagnosis and overtreatment. However, the situation is changing, and three statistical approaches have recently been proposed. The first is a probability model based on the chances of success per cycle and suggests for the definition three failed oocyte retrieval cycles with all embryos being transferred in women younger than 40 years of age. The second approach suggests an individualized diagnosis that takes into consideration multiple factors, while the third is also based on individualization but mainly relies on anticipated euploidy rates across the female age range. All these approaches have their pros and cons. Regardless of the specific peculiarities, they represent steps in the right direction, with the intent of providing a statistically sound definition. However, these attempts will not be useful unless endorsed by the scientific community in general. There is a pressing need for a rigorous and shared definition of RIF that will be widely accepted by researchers, scientific societies and other stakeholders, including patients.