Over the past 20 years, a large number of studies have been carried out to determine the mechanical properties of fibres of all kinds, particularly in tension and for various applications. For plant fibres, 3 types of tests are commonly used: tensile tests on single/elementary fibre, tensile tests on bundle of fibres, and the inverse method IFBT (Individual Fibre Bundle Testing). Most studies focus on the use of tensile tests on elementary fibres. Despite its advantages, this method is time-consuming and challenging because of the tiny size of the fibres and the many sources of uncertainty. The aim of the present interlaboratory experimental campaign, involving 9 research groups, was to gain a better understanding and quantify the sources of variability when estimating the tensile properties of plant fibres. For that purpose, three batches of fibres were selected: two plant fibres, i.e. combed hemp and flax fibres as well as a synthetic organic fibre, i.e. aramid fibres, for benchmark purpose. A total of approximately 1250 fibres were tested. Results highlight significant intra- and inter-laboratory variability. For aramid fibres, median values of initial apparent tangent tensile moduli Ei, tensile strength σR, and strain at failure εR fell within the ranges of 68-97 GPa, 2821-3981 MPa and 3.4% to 4.5%, respectively. For flax, the values were in the ranges of 10-51 GPa, 383-1107 MPa, and 1.2% to 2.9%. For hemp, the ranges were 16-33 GPa, 252-577 MPa, and 1.4% to 3%. Furthermore, the Coefficients of Variation (CoV) ranged from 8% to 22% for Ei, 10-23% for σR, and 8-23% for εR for aramid fibres. The CoV were significantly higher for the tested plant fibres with 28-54%, 32-59%, and 23-44% for flax and 38-68%, 51-80%, and 36-58% for hemp. Human factors and experimental procedures, especially for the estimation of the fibre cross-sectional area and the tensile strain, were identified as the main sources of scatter in tensile properties. Post-processing procedures, particularly determining the starting point of the tensile test, are crucial, involving the elimination of slacks in fibre and load trains. The more pronounced scatter for flax and hemp compared to aramid was attributed to the difficulty of analysing the high variability and complexity of plant fibre cross-section and the greater diversity of non-linearities observed on stress-strain curves. The main recommendations for reducing dispersion in tensile testing include considering geometric models adapted to fibre morphology when determining the cross-sectional area, maintaining precisely controlled hygrothermal conditions, accurately defining zero load and zero displacement points, and systematically specifying the strain range used to determine apparent tangent modulus. Finally, alternative methods such as Digital Image Correlation should be considered to reduce uncertainties in strain measurements.