Field estimates of net precipitation in primary evergreen forest in Ivory Coast, West Africa, are compared with estimates given by two models: the Gash model, which assumes all rain on a single day to have fallen in one shower, and the Mulder model, which uses a more realistic, equalized rainfall distribution. Metereological data were collected in a clearing of about two hectares. Throughfall, measured using a roving gauge technique, was 90.8% of gross rainfall (1022 mm in five months). Both models simulated this figure well within the error range of the field estimates. Errors in the model predictions are thought to result from errors in the parameter estimates, and, for the Gash model, from the assumption that daily rainfall comes in one shower. It is suggested that a model combining the advantages of both might be suitable for routine application in the humid tropics.