This article considers errors of care in neonatology. In the 19th century errors that resulted in high infant mortality were shaped by the social environment, and in this setting the development of the incubator failed. In the early 20th century, with the emergence of the modern hospital as a technological, science-driven system, physicians had more control of patients' environments, and thus medical errors could occur from systematic care and affected larger numbers. Later in the 20th century, the development of randomized controlled trials and systematic reviews began to improve care and to decrease the risks associated with new treatment methods. Large variations in practice still exist between physicians as individuals and institutions. Considering these variations as risks has led to the use of institutional databases, benchmarking and clinical care guidelines. The efficacy and safety of these methods is unproven. Risks will never disappear from medicine. The question of what risks are 'acceptable' is, in general, unanswerable.