Three census methods for studying fruiting phenology and fruit production were compared: platform observation, fruit-trap collection and raked-ground survey. The estimation of fresh fruit mass varied greatly with the census method used. Moreover, the fruiting phenology recorded using the fruit-trap collection and raked-ground survey methods showed a delay of one to two months compared with the platform observation method. Any of the three methods can be used for the approximation of annual fruiting phenology, but that of fruit-trap collection is suited to measure the annual fruit production, and that of platform observation offers the best estimation of the quantity of fruit in tree crowns at a given time.