Quantifying species relative abundances in plant communities remains a key issue for the assessment of community functional structure. This is particularly challenging when non-destructive estimates are required over time. We tested whether the point intercept method (PIM), originally developed for low-diverse communities, is relevant for assessing the aboveground biomass and functional structure of highly diverse, low-productive Mediterranean grasslands. We sampled 18 communities with the PIM along a gradient of soil depth and texture, twice over the growing season. After each sampling period, we harvested the aboveground biomass in order to measure species biomass and to assess species richness and community functional structure with plant height, leaf area and leaf dry matter content (LDMC). We investigated the relationship between point intercept measurements and aboveground biomass at three hierarchical levels (species, growth-form and community) to find generalizable calibration equations for estimating community biomass and tested for sensitivity of estimates to community structure. We then compared the community weighted mean (CWM) and variance (CWV) of LDMC, calculated with and without calibration. Differences in species growth strategy and phenology strongly impacted biomass estimates at both the species and the community level. These differences were, however, successfully accounted for by growth-form specific calibrations, which provided accurate estimates without any influence of community structure. Lack of calibration may have dramatic consequences on functional structure assessment by inducing errors in estimates of CWV up to 80 %, depending on growth-form proportions. This work contributes to a better understanding of the possible methodological biases induced during sampling with the PIM, when quantifying species relative abundances for functional structure assessment in complex communities.