OBJECTIVE To determine the effectiveness of a single dose of prednisone administered by a parent to a child early in an asthma attack. METHODS A randomized, double-blind, placebo-controlled, crossover study with children enrolled for 12 months (6 months prednisone, 6 months placebo). METHODS A primary-care clinic and emergency department of an inner-city teaching hospital from March 1992 through May 1993. Children 2 to 14 years of age enrolled in this clinic who had made two or more outpatient (emergency department or primary-care clinic) visits for acute asthma in the preceding year. METHODS There were 204 eligible children, of whom 86 were contacted and enrolled; of these, 78 (91%) completed the study. METHODS Capsules containing prednisone (2 mg/kg up to 60 mg) or placebo. Parents were instructed to give their child one capsule for an asthma attack that had not improved after a dose of the child's regular acute asthma medicine. METHODS Parents were interviewed every 3 months. Computerized patient records and chart reviews were used to verify parent reports. Outcome measures were the numbers of outpatient visits and hospitalizations for treatment of acute asthma. RESULTS Neither the total number of attacks nor the number for which medicine was used differed significantly by arm of study. There was a larger number of attacks resulting in outpatient visits when children were in the group that received prednisone (1.1 +/- 0.59 versus 0.59 +/- 0.86). This trend was less pronounced but persisted when limited to attacks for which the medicine was given (0.58 +/- 0.99 versus 0.35 +/- 0.55). Neither the number of attacks resulting in admission nor the number of hospital days differed significantly by arm of study. CONCLUSIONS A single dose of prednisone available for use at home early in an asthma attack was associated with an increase in outpatient visits made for acute asthma. When prednisone was given for an attack, there was no reduction in outpatient visits. This intervention can not be recommended for children with asthma. These results should be confirmed in other pediatric populations.