Despite impressive recent advances in neonatology, outcomes for extremely premature, very-low-birth-weight infants (500 to 750 g) remain uneven. In a situation of inherent uncertainty, treating patients vigorously could do violence to the moral principles of nonmaleficence and (distributive) justice. Equally, failing to treat patients vigorously because of concerns about nonmaleficence and (distributive) justice could violate the principle of patient-centered beneficence. Compounding this dilemma is the legacy of the "Baby Doe Regulations." International perspectives on this particular quandary are provided. We assert that at Stanford (Calif) University the "individualized prognostic strategy" rather than the "wait until certainty" approach prevails. Four concluding questions are posed: Why is prevention not encouraged more than after-the-fact heroic intervention? Is it possible to develop a more rational view of stopping aggressive therapy once having started? Can we ignore the finitude of our medical resources? Is there a need to redefine the nature of autonomy?