Each flower in a plot of the red raspberry cv. Glen Isla was identified by attaching a date-coded tag to its pedicel at anthesis. The berries were then machine harvested on 4 occasions during the ripening period. An Agricultural Sciences Inc. harvester, travelling at 1.6 km/h with a finger vibration frequency of 8.3 Hz and a finger stroke of 6.5 cm, was used. An analysis (by number) of the fate of those berries which were available to the harvester and would have ripened by the time of the final harvest showed that 12% of the crop was shed pre- and inter-harvest, 73% was removed in a mature and 5% in an unripe condition, 5% was on broken laterals and a further 5% was left on the plants after the final harvest. The results of the present experiment were used to explain why 40% of the potential crop could not be accounted for when an earlier experiment [see HcA 45, 9330] was machine harvested. If allowances were made for (i) a lower mean berry weight of the unripe berries when calculating the weight of fruit they would have produced if they had been allowed to ripen, (ii) berries below the height of the machine's catching plates and therefore not available to the harvester, (iii) pre- and inter-harvest shedding of over-ripe fruit and (iv) berries on broken laterals, an additional 30% of the crop could be accounted for. The remaining 10% loss was attributed to the effects of mechanical damage to the plants and/or the developing berries, leading to a reduction in berry size and therefore in fruit yield. It is suggested that both berry weight and number must be recorded when comparing different methods of harvesting, so that any changes in berry size can be identified.