Illogically, and with no empirical evidence to support it, a mischievous view has been promoted that the interest of the vast number of patients involved in the poorly controlled experiments of informal medical ‘tinkering’ are less in need of protection than are those of the relatively small number of patients who are involved in planned, properly controlled clinical experiments.1

What are the responsibilities of paediatricians and other clinicians when there is uncertainty about whether particular treatments are more likely to do good than harm, or vice versa? The General Medical Council's advice to doctors is clear: ‘You must … help to resolve uncertainties about the effects of treatments’. The trouble is that circumstances are stacked against doctors who aspire to this standard. As the British paediatrician Richard Smithells noted decades ago, ‘I need permission to give a drug to half of my patients, …