Large monitoring programs exist in many countries and are necessary to assess present and past biodiversity status and to evaluate the consequences of habitat degradation or destruction. Using such an extensive data set of the floristic richness in the Paris Ile-de-France region (France), we compared different sampling efforts and protocols in different habitat units to highlight the best methods for assessing the actual plant biodiversity. Our results indicate that existing data can be used for a general understanding of site differences, but analysts should be aware of the limitations of the data due to non-random selection of sites, inconsistent observer knowledge, and inconsistent sampling period. The average species diversity recorded in a specific habitat does not necessarily reflect its actual diversity, unless the monitoring effort was very strong. Overall, increasing the sampling effort in a given region allows improvement of the (1) number of habitats visited, (2) the total sampled area for a given habitat type, (3) the number of seasons investigated. Our results indicate that the sampling effort should be planned with respect to these functional, spatial and temporal heterogeneities, and to the question examined. While the effort should be applied to as many habitats as possible for the purpose of capturing a large proportion of regional diversity, or comparing different regions, inventories should be conducted in different seasons for the purpose of comparing species richness in different habitats.