Data from 87 trees within 21 natural infestations were used to estimate the population size of the southern pine beetle (Dendroctonus flontalis Zimmermann) under collapsing high-level conditions in North Carolina. Spots of K trees (1<K<12) were created by random selection without replacement from the initial pool. Samples of m trees (1<m<K) were then chosen from each simulated spot according to random and selective plans. These samples over m trees were then extrapolated to the full K tree spot by 1 of 3 scaling factors: percentage of trees sampled, percentage of diameter sampled, or percentage of phloem area sampled. In this manner the total numbers of parent adults, brood, and brood adults were estimated for each artificial spot. Proportional errors, defined as (observed value – expected value)/(observed value), were computed over 10 000 independent trials for a variety of sample sizes, spot sizes, and scaling factors. The precision of these errors, after a bias correction was applied, was used to compare the different schemes. Random-diameter and largest-diameter schemes were judged best because of their good precisions and ease of application. However, when these results were applied to the 21 natural infestations, the confidence intervals were found to be too narrow to include the expected number of estimates. A discussion of this shortcoming is discussed in relation to spot dynamics and its effect on the application of this sampling technique.