This research aimed at testing different plot and sample sizes, as well the sampling design to evaluate diversity indexes in a natural forest. The study was carried out in the Irati National Forest, Paraná. A combination of sampling schemes resulted in 28 treatments, which were evaluated considering the number of species, the number of sampled individuals and calculation of the Margalef, Shannon and Simpson indexes. The species accumulation curve was carried out based on three approaches: random sequential numerical order, ascending order of species number in each plot, and order of the parcel with the highest number of species for the smaller number of species. The random sample design represented at best the number of species in smaller areas (richness), and the systematic sample design represented better the biggest sampled areas. The Simpson index and Shannon index were not sensitive to represent the variations found in relation to the census. The Margalef index had better sensitivity to express species diversity of the forested area. The species accumulation curve proved its feasibility to represent a favorable and sufficient sampling size.