A literature search of personal files, the Oxford Database of Perinatal Trials, Medline, and the Research Update databases, identified 19 perinatal trials published between October 1991 and September 1992. Included were only those trials in which the efficacy or effectiveness of an antenatal intervention was assessed by a neonatal outcome. A quality assessment of each study was performed. Quality scores ranged from 0.25 to 0.80 (maximum possible score, 1.00). Most trials did not show a statistically significant change in the primary outcome measure, because of an overestimate of the treatment effect on which the sample size had been based. Thus, clinically important treatment effects often may be overlooked. None of these trials on their own showed beyond doubt whether the intervention under study did more good than harm. There is a need to base sample size calculations on more realistic, but still clinically important, event rate differences to show the effectiveness of perinatal interventions aimed at improving neonatal outcomes. To do so usually requires a large sample size, necessitating multicenter trials with national and international cooperation.