An early recommended control measure for cocoa pod borer (CBP) in Sabah was frequent thorough harvesting to a low standard of ripeness, here called regular complete harvesting (RCH). This reduces the opportunity for CPB to complete its development. Chemical sprays were also recommended, commonly five applications of BHC or a synthetic pyrethroid at 10 day intervals, during the period of trough crop, or recently, spraying in response to infestation levels. This paper reports a trial that demonstrated the effect of RCH. Economic control is achieved if 50% or fewer pods are infested and this can easily be achieved by RCH. A late commercial‐scale trial confirmed that it could bring infestation to much lower levels, whilst spraying made little difference. The trials assessed spraying in response to CPB thresholds. Initially spray was in response to moth capture in pheromone traps (January‐June 1986), later to pod infestation level. The field was split into two, with a low infestation threshold in one part (L), and a high one in the other (H). Infestation followed the expected pattern, rising in crop troughs (July to September), and declining in peaks. Despite frequent spraying in L area and very little in H, infestations were similar. In 1988, when RCH was stressed, infestation started low in both areas, crossing only the L threshold, on only two occasions. Other investigations confirmed increasing infestation towards ripeness (when colour change begins). Egg‐laying and subsequent emergence can continue from unharvested ripe pods. Practical difficulties for RCH include low harvester incomes in the trough, and a reluctance to harvest infested pods. RCH gives high cost per unit product in the trough, but this is counteracted by extra crop. RCH is a pest management technique without adverse side‐effects. Further work is needed on the economics of RCH strategies, including abandoning harvesting in trough periods, monitoring of harvesting efficiency and the critical number of unharvested pods compatible with satisfactory control.