Productivity is a primordial selection criterion for new varieties of cocoa. Evaluation at the research station by the usual methods is restrictive and requires a lot of work. The aim of this work was to evaluate the efficiency of simple methods which are applicable to cocoa selection. To do this, the "classical" method of harvesting from individual trees (based on counting and weighing fruits on a monthly basis) has been compared with counting and visually scoring, on a scale of 0 to 5, of the quantity of fruits on the tree before the main harvest and before the mid-crop during two agricultural years (Septemeber 2002 to June 2004). The number of peduncles and fruit-bearing cushions were also counted and visually scored at the end of the one campaign (June 2003). Black pod infection level was estimated with the same methods as applied for pod yield during one harvesting campaign (2003/04). Highly significant correlations have been established between the first three methods to assess pod yield (i.e. "classical" method, counting and scoring the number of fruits). The use of counting and especially scoring the pod load before the main harvesting periods would make it possible to significantly reduce staff requirements and to still be able to obtain good quality information, useful for selection of individual trees and hybrid families. Counting and scoring of rotten pods before harvest campaigns seemed to be informative for the actual degree of infection during the harvest campaign, especially at the level of mean family values. On the other hand, the simplified method is less well correlated with the cumulated degree of infection over several years. The counting and scoring of active flower cushions at the end of the mid crop (Methods 4 and 5) appeared also to be promising for estimating average family pod production for the same harvesting campaign and for cumulated pod production (r = 0.78 to 0.89). However, the correlation at individual tree level was much lower (r