Canopy arthropods were sampled by insecticide fogging to study their community structure in two New Caledonian primary rain forests. The representativeness of these samples was analysed by two different methods: the diversity-area relationship and the relationship between the distribution of the taxa and the sample size, using Pielou's method. The results showed that the higher the degree of aggregation, the higher must be the minimum sample size to ensure a stable distribution. In the same way, the higher the diversity index, the higher must be the sample size to ensure a representative sample of the community. In this study, 40 sample units of 1 m2 were used, although, samples of 9 to 25 m2 seem to be sufficient according to the distribution of the taxa sampled. Five to 30 m2 should be sufficient to ensure representative samples of the whole community for estimating diversity.