To test the hypothesis that codeine and dextromethorphan are effective in alleviating the symptoms of acute cough, we conducted a randomized, controlled trial. Eligible patients were children 18 months to 12 years of age, seen in private pediatric practices, with significant night cough of less than 14 days' duration. Study patients were randomly selected to receive codeine, dextromethorphan, or placebo at bedtime for 3 consecutive nights. Outcomes were assessed by the use of a parent questionnaire rating the severity of symptoms at the initiation of therapy, and after each night of the study. Every patient had a cough score (range 0 to 4) and composite symptom score (range 0 to 9) computed for each day of the study. One hundred forty-one doses of study medication were evaluated in 49 patients, including 13 children receiving placebo, 19 dextromethorphan, and 17 codeine. Mean cough and composite symptom scores decreased in each of the three treatment groups on each day of the study; there were no significant differences. Regression analysis, with reduction in cough score as the outcome of interest, showed that neither dextromethorphan nor codeine was significantly more effective than placebo (p = 0.41 and 0.70, respectively). Reduction in cough score was positively correlated with the severity of cough at the start of treatment (p = 0.007). Our data suggest that, in the doses used, neither codeine nor dextromethorphan is superior to placebo in treating night cough in children.