Expected incidence of acute malnutrition is the most appropriate measure for projecting the needs of a nutritional treatment program over time in terms of staffing, food, and other treatments, but direct estimation of incidence is rarely feasible at the onset of an intervention. While incidence may be approximated as prevalence/average duration, ethical constraints preclude measurement of the duration of acute malnutrition in the absence of treatment. The authors used a compartmental model to estimate the duration of untreated moderate acute malnutrition (MAM) and severe acute malnutrition (SAM) in children aged 6-60 months. The model was informed by data from a community-based cohort of children in Niger followed from August 2006 to March 2007. Maximum likelihood estimates for the duration of untreated MAM, defined by weight-for-height z score and middle upper arm circumference, were 75-81 days and 101-116 days, respectively. The duration of untreated SAM, defined by weight-for-height z score, was 45 days. The duration of untreated MAM appears to have been shorter among children aged 6-35 months compared with those aged 36-60 months. Such estimates of the duration, and thus incidence, of untreated malnutrition can be used to improve projections of program needs and estimates of the global burden of acute malnutrition.