Objective/Hypothesis: Data from cohort studies and untreated groups in randomized controlled trials can be identified through systematic literature review and synthesized with meta-analysis to estimate natural history of acute otitis media (AOM) and otitis media with effusion (OME). Study Design: Systematic literature review and meta-analysis. Method: Source articles were identified by MEDLINE search through August 2002 plus manual crosschecks of bibliographies and published meta-analyses. Data were abstracted independently by two investigators and combined with random effects meta-analysis to estimate spontaneous resolution, 95% confidence intervals (CI), and heterogeneity. Sensitivity analysis was performed. Results: Sixty-three articles met inclusion criteria. AOM symptoms improved within 24 hours without antibiotics in 61% of children (95% CI, 50–72%), rising to 80% by 2 to 3 days (95% CI, 69–90%). Suppurative complications were comparable if antibiotics were withheld (0.12%) or provided (0.24%). Children entered recurrent AOM trials with a mean rate of 5.5 or more annual episodes but averaged only 2.8 annual episodes while on placebo (95% CI, 2.2–3.4). No AOM episodes occurred in 41%, and only 17% remained otitis prone (3 or more episodes). OME after untreated AOM had 59% resolution by 1 month (95% CI, 50–68%) and 74% resolution by 3 months (95% CI, 68–80%). OME of unknown duration had 28% spontaneous resolution by 3 months (95%, CI 14–41%), rising to 42% by 6 months (95% CI, 35–49%). In contrast, chronic OME had only 26% resolution by 6 months and 33% resolution by 1 year. Conclusions: The natural history of otitis media is very favorable. Combined estimates of spontaneous resolution provide a benchmark against which to judge new or established interventions. The need for surgery in children with recurrent AOM or chronic OME should be balanced against the likelihood of timely spontaneous resolution and the potential risk of learning, language, or other adverse sequelae from persistent middle ear effusion. Further research is needed to identify prognostic factors that can target children unlikely to improve spontaneously for earlier intervention.