In Brazil, urea is the most used nitrogen (N) fertilizer to improve forage production. However, their excessive use can cause environmental impacts through N losses, such as ammonia (NH3) volatilization. Therefore, the current study adjusted and estimated the NH3 volatilization from urea applied on tropical pastures in three rainfall conditions using mathematical models. Data were collected from Marandu grass (Brachiaria brizantha) fertilized with 50 kg N ha-1 during wet, intermediate, and dry conditions. Ammonia volatilization was measured in five semi-open chambers for 21 days. The linear, quadratic, exponential, Gompertz, Groot, and Richards models were tested for fitting and estimating the NH3 volatilization. The Gompertz, Groot, and Richards models generated predictions similar to the observed data, with a high determination coefficient, indicating a better fit of these equations to data, with precision and accuracy. However, the Groot model was selected due to the lowest root mean square error of prediction (0.29 % total N lost as NH3). The greatest N loss as NH3 volatilization occurred in the wet, followed by intermediate and dry conditions (20.2, 17.0, and 11.3 % total N lost as NH3, respectively). Therefore, nitrogen losses as NH3 volatilization after application of 50 kg N ha-1, as urea source, are altered according to the weather conditions, reaching 20% of N added in the wet rainfall period. The Groot model is recommended for fitting and estimating the NH3 volatilization from urea applied on Marandu grass pastures in the wet and dry rainfall conditions.