Litter fall collection is a direct method widely used to estimate leaf area index (LAI) in broad-leaved forest stands. Indirect measurements using radiation transmittance and gap fraction theory are often compared and calibrated against litter fall, which is considered as a reference method, but few studies address the question of litter specific leaf area (SLA) measurement and variability. SLA (leaf area per unit of dry weight, m 2 ·g -1 ) is used to convert dry leaf litter biomass (g·m -2 ) into leaf area per ground unit area (m 2 ·m -2 ). We paid special attention to this parameter in two young beech stands (dense and thinned) in northeastern France. The variability of both canopy (closure, LAI) and site conditions (soil properties, vegetation) was investigated as potential contributing factors to beech SLA variability. A systematic description of soil and floristic composition was performed and three types of soil were identified. Ellenberg's indicator values were averaged for each plot to assess nitrogen soil content. SLA of beech litter was measured three times during the fall in 23 plots in the stands (40 ha). Litter was collected bimonthly in square-shaped traps (0.5 m 2 ) and dried. Before drying, 30 leaves per plot and for each date were sampled, and leaf length, width, and area were measured with the help of a LI-COR areameter. SLA was calculated as the ratio of cumulated leaf area to total dry weight of the 30 leaves. Leaves characteristics per plot were averaged for the three dates of litter collection. Plant area index (PAI), estimated using the LAI-2000 plant canopy analyser and considering only the upper three rings, ranged from 2.9 to 8.1. Specific leaf area of beech litter was also highly different from one plot to the other, ranging from 150 to 320 cm 2 ·g -1 . Nevertheless, no relationship was found between SLA and stand canopy closure or PAI. On the contrary, a significant relationship between SLA and soil properties was observed. Both SLA and leaf area had the lowest values in the most hydromorphic soil with the highest nitrogen content. On the other hand, the highest values of SLA and leaf area were observed on the plots with the lowest nitrogen content. This spatial variability of SLA was taken into account to estimate LAI from litter collected at eight plots. For our study site, we conclude that neglecting SLA spatial variability is at the source of 8-24% error in the calculation of LAI.