Species richness is a key variable in measuring diversity of ecological communities. It is crucial to get reliable estimates for the number of plant species in space (mapping) and - even more important in the context of monitoring - over time. Therefore, knowledge on error rates related to recordings of species numbers should be considered in such inventories. The performance of observers in four field tests to capture species numbers carried out in forest ecosystems in central and southern Europe were compared. Observer-related species accumulation (rarefaction) curves and derived efficiency curves were analysed, resulting in mean error rates of 29.7% and 39.4% over series of plots sized 4m(2) and 100m(2) respectively. As a new approach individual rarefaction and efficiency curves reveal site-specific and spatially differentiated capabilities of observers to register plant species. Since expertise and individual searching strategies are difficult to parametrise, reasons for variation in error rates remain largely unknown. However, statistical modelling with site- and scale-specific mean error rates gave an overview on important influential factors like location, scale, spatial integration, and their interactions. Our results underline the importance to incorporate specific training and inter-comparison measures in monitoring programs and critical perception of results on temporal changes of species richness.