Objective: This systematic review and meta-analysis examine the contribution of duration to treatment outcome for eating disorders. Method: Studies (n = 31) were identified that examined associations (r) between duration and 45 different outcomes. We were unable to extract r for seven studies (9 outcomes) and extracted r for 36 outcomes across 24 studies (2,349 participants). Indicators of treatment outcome were heterogeneous and thus a series of different meta-analyses, aimed at increasing homogeneity, were conducted. Results: First, we examined the average effect size for one primary eating disorder related outcome from each of the 24 studies. There was no association between duration and treatment outcome (r = .05, 95% CI: −.03:.13), with high heterogeneity. Second, we conducted three subgroup analyses to explore possible sources of heterogeneity (diagnosis: anorexia nervosa versus bulimia nervosa; nature of the outcome: binary versus continuous; or type of outcome: binary indicator of recovery, eating disorder psychopathology, weight gain). There was no significant moderation or associations between duration and outcome (ranging from .02-.08), with low to medium heterogeneity. Third, two stand-alone analyses examined outcomes related to weight gain (n = 8) and eating disorder psychopathology (n = 5), with nonsignificant rs of .23/−.06, respectively. High levels of heterogeneity were present. Discussion: Duration did not influence treatment outcome across any of our metaanalyses. Increasing homogeneity and power will allow more stable estimates of the impact of duration on outcome to be calculated; to this end, future treatment studies should include outcome related to weight gain (anorexia nervosa) and improvements in eating disorder psychopathology. Resumen Antecedentes: Esta revisión sistemática y metaanálisis examinan la contribución de la duración al resultado del tratamiento para los trastornos de la conducta alimentaria. Métodos: Se identificaron estudios (n = 31) que examinaron las asociaciones (r) entre la duración y 45 diferentes resultados. No fue posible extraer r para siete estudios (nueve resultados) y se extrajo r para 36 resultados en 24 estudios (2349 participantes). Los indicadores del resultado del tratamiento fueron heterogéneos y, por lo