To set the foundation to develop a disease-based, operation-specific model to predict the outcome of pediatric airway reconstruction surgery, we performed a retrospective database review of children operated on at a single, tertiary-care children's hospital. Over the 12-year period 1988 to 2000, a total of 1,296 airway reconstruction procedures were performed. Out of these, charts were identified for 199 children who underwent laryngotracheal reconstruction for a sole diagnosis of subglottic stenosis. Children were excluded from the study if their disorder included supraglottic, glottic, or upper tracheal disease. The main outcome measures were Myer-Cotton grade-specific decannulation and extubation rates, including both operation-specific and overall results. There were 101 children who underwent double-stage laryngotracheal reconstruction. The operation-specific decannulation rates for Myer-Cotton grades 2, 3, and 4 were 85% (18/21), 37% (23/61), and 50% (7/14) (chi2 analysis, p = .0007). The overall decannulation rates were 95% (20/21), 74% (45/61), and 86% (12/14) (chi2 analysis, p = .04). There were 98 children who underwent single-stage laryngotracheal reconstruction. The operation-specific extubation rates for Myer-Cotton grades 2, 3, and 4 were 82% (37/45), 79% (34/43), and 67% (2/3) (chi2 analysis, p = .63). The overall extubation rates were 100% (45/45), 86% (37/43), and 100% (3/3) (chi2 analysis, p = .03). Logistic regression analysis showed no effect of age (less than or greater than 2 years of age) on operation-specific or overall outcome parameters. We conclude that laryngotracheal reconstruction for pediatric subglottic stenosis remains a challenging set of procedures in which multiple operations may be required to achieve eventual extubation or decannulation. Children with Myer-Cotton grade 3 or 4 disease continue to represent a significant challenge, and refinements of techniques are being examined to address this subset of children. Disease-based, operation-specific outcome statistics are the first step in the development of a meaningful predictive model.