Screening for asymptomatic disease is a routine aspect of contemporary public health practice. However, it is also controversial, because it leads to overdiagnosis and overtreatment, with many arguing that programmes are “ineffective”, i.e. the “costs” outweigh the “benefits”. This paper explores a more fundamental objection to screening programmes: that, even if they are effective, they are ethically impermissible because they breach the principle of non-maleficence. In so doing, it suggests a new approach to the ethics of risk, justifying a concern with how policies affect individuals’ absolute ex-ante prospects. Part 1 sets up the tension between screening and non-maleficence. Part 2 introduces and motivates a novel interpretation of the non-maleficence principle, “ex-ante Do No Harm”, which resolves this tension. Part 3 defends and clarifies this principle by discussing its relationship to the ex-ante Pareto principle. Part 4 discusses the worry that risk estimates are too “subjective”.