A prospective, double-blind crossover study design was used in the evaluation of the role of prednisone in the treatment of a chronic middle ear effusion. Sixty children with an effusion--of at least 2 months' duration--randomly received either prednisone or placebo for 2 weeks. One week later they were re-evaluated. If the effusion had resolved, they were followed periodically; if the effusion persisted, they were given the crossover regimen and reevaluated. Sixty percent of the patients' effusions cleared, but there was no statistical difference between the prednisone and placebo groups. The amount of hearing recovery was also the same for both treatment regimens. Subdividing the patients on the basis of effusion duration, unilateral vs. bilateral effusions, or history of ventilation tubes or antibiotics failed to identify a subgroup of individuals who might benefit from a short-term course of oral steroids.