Although theophylline is a widely used drug for the treatment of acute childhood asthma, its efficacy has not been clearly established. This study constitutes a meta-analysis of published randomized clinical trials of theophylline in children hospitalized with acute asthma. We conducted a search of English language MEDLINE citations from 1966 to 1995 and analyzed the methods of each report meeting study criteria. We pooled similar clinical measures across studies if a test for homogeneity of effect size was non-significant. The six methodologically acceptable randomized clinical trials included a total of 164 children less than 18 years of age. Incomplete reporting of measures and variances was common. No study included children in intensive care settings. Using pooled results, pulmonary function parameters [forced expired volume in 1 second (FEV1), forced expired flow (FEF)] appeared better at 24 hours in the theophylline group, but the results did not reach statistical significance (mean effect difference, + 3.9% predicted values; pooled effect size, + 1.6 SDS; P = 0.25). A mean of 2.1 more albuterol treatments were administered in the theophylline group (pooled effect size, - 0.18 SDS; P = 0.02), and the mean hospital stay was slightly longer (mean effect difference, - 0.31 days; pooled effect size, - 0.18 SDS; P = 0.03). We conclude that currently available data do not indicate a significant beneficial effect of theophylline in children hospitalized with acute asthma. There is evidence for weak detrimental effects. Theophylline efficacy in intensive care unit settings remains unstudied.