The effects of several elements of floristic structure and bioindication study design on the predictive power of the Index of Atmospheric Purity (IAP) proposed by LeBlanc & De Sloover were studied using Monte Carlo simulation. The predictive power was assessed by the correlation coefficient of the IAP value, computed under specified assumptions, and the stand position along a simulated pollution gradient. The effects of seven parameters were studied: proportion of sensitive and poleophilous species, species frequency distribution, size of total species pool, number of stands, distribution of stands along the air pollution gradient and index choice. The simulations showed that the median predictive power of such studies is affected mainly by the proportion of sensitive and poleophilous species, the number of species, and distribution of trees along the pollution gradient. The variance of the predictive power is influenced mainly by the proportion of sensitive species, frequency distribution, number of stands and distribution of stands along the pollution gradient. Recommendations for further bioindication studiesare given.