Population densities of highly aggregated insect species in complex and variable habitats can be difficult to estimate efficiently, accurately, and with minimal variance. Aphis pomi DeGeer exemplifies this situation, in that its infestations typically are found near the tips of vigorously growing shoots of apple, Malus domestica Borkh, which vary in number and distribution throughout the growing season. Because A. pomi populations are permanently confined to vigorous shoots, randomly selecting shoots to sample is inefficient and unnecessarily increases the variance in population estimates. Previous workers resolved many difficulties in sampling A. pomi by recommending the use of the number of aphids on the most infested leaf as a reliable predictor for the number of aphids on the shoot. While this sampling protocol has been widely adopted, it is most applicable early in the growing season or in other unusual situations when tree vigor is high and randomly selected shoots are likely to be infested. In this paper, we confirm that the most infested leaf is still a reliable predictor for the number of aphids on a shoot throughout the growing season, even when A. pomi densities are very low. However, we also suggest that the efficiency and performance of the most infested leaf as a predictor can be improved by using preliminary sampling to determine the proportion of vigorous shoots in the orchard or block of interest and sampling for A. pomi only from those vigorous shoots.