With the full survey data for a 24-ha subtropical evergreen broad-leaved forest dynamics plot, we evaluated spatial variation in forest structure characteristics (basal area and aboveground biomass), and calculated the minimal sample size and total sampling area necessary to estimate the forest structure characteristics within 20% (+/- 10%) of the observed values with 95% probability for particular quadrat sizes by using a computer program that is designed to simulate the sampling process by allowing different sized quadrats to be randomly located within the sampling region. We found that (1) based on the 600 20 mx20 m subplots, basal area and aboveground biomass displayed a high degree of variation, with respective coefficients of variation of 27% and 31%; (2) based on the computer simulation analysis, the variability of basal area and aboveground biomass decreased with increasing quadrat size. The number of quadrats required to achieve the specified degree of precision dropped sharply with the increase of quadrat size. However, the total sampling area increased with increasing quadrat size, suggesting that using several small quadrats across the sampling area is more efficient than using fewer larger quadrats. Results of this study are valuable for evaluating the reliability of previous research and may assist researchers in designing effective sampling strategies for future field surveys, particularly in subtropical evergreen broad-leaved forests in China.