Leaf chlorophyll content (LCC) is an important indicator of plant health. Earth observation facilitates LCC monitoring on large spatial scales and within short time intervals by retrieving canopy LCC from spectral signals. However, in-situ measurements are still necessary to validate the predicted values and the accurate quantification of LCC is labour-intensive and costly. Handheld leaf chlorophyll metres, such as the SPAD-502 (Konica Minolta) promise quick and non-destructive in-situ LCC estimates. SPAD values can be easily converted to LCC by established calibration curves that are based on measurements of single plant species. Due to these practical advantages, the application of SPAD-LCC calibration curves, originally derived in an agricultural context (Markwell et al., 1995), has become common practice in field studies. Although associated with a variety of uncertainties, the transfer of general SPAD-to-LCC calibration equations to natural ecosystems has not yet been comprehensively investigated. In this study, we test the reliability of SPAD-to-LCC conversions in three structurally different semi-natural grassland areas with mixed species compositions. We measured SPAD values in the field and determined the corresponding LCC of the very leaf samples via chemical extraction and spectrophotometry. Our results show that SPAD-to-LCC conversion equations lack accuracy in semi-natural grassland with mixed species compositions (n = 157, R2 = 0.046, p ¡ 0.05, for all sites). This study highlights the uncertainties in optical approaches for in-situ LCC assessments. It further underlines the need to find feasible and reliable solutions for a quick and efficient chlorophyll determination in natural habitats with mixed species compositions towards a reliable monitoring of ecosystem properties.