A water-film above the leaf surface is a necessary condition for the start of the infective process of many pathogens. Therefore, the Leaf Wetness Duration (LWD) has a strong relationship with the development and outbreak of plant diseases. The occurrence of free water on leaf surfaces is a common situation for the simple plastic greenhouses used to grow ornamental crops in the Bogota Plateau (Colombia). The objective of this work was to evaluate the suitability of four models to predict LWD occurrence in a greenhouse rose crop. Model performances were compared against the measurement of dielectric leaf wetness sensors, placed at 1.2 and 1.8 m above the ground, within two sampling locations inside a greenhouse during a 25-day period. Sensors were connected to dataloggers, programmed to store data every 10 min. Two copper-constantan thermocouples measured air and wet-bulb temperature and were used to calculate relative humidity (RH). Three empirical models were evaluated: Constant RH threshold (RH≥90%), locally calibrated RH threshold (RH≥94%) and dew point depression (DPD). Also, one physical model that estimates LWD based on the latent heat flux (LE) of the leaf was considered. As a result the following four scores were calculated: fraction of correct estimates, correct success index, false alarm ratio and bias. LWD estimated at 1.8 m above the ground showed the best performance for all the empirical models whereas the physical model yielded the best results for measurements made at 1.2 m. The results obtained with the models indicated a differential degree of success for the prediction of LWD. In general, RH and DPD with thresholds of 94% and 2°C respectively were the most suitable models to estimate LWD, resulting in higher precision and accuracy. The results will contribute to the development of integrated pathogen management in greenhouse rose crops in the Bogota Plateau.