Natural fibres obtained from plant sources are attractive as a replacement for glass fibres in fibre reinforced plastic composites because of their environmental benefits. However, unlike synthetic fibres, natural plant fibres show considerable variability in their mechanical properties due to the effects of climate, soil quality, time of harvest, etc. Variability in properties of the fibres translates into variability in the properties of products made from natural fibre composites and this is a major obstacle to the more widespread use of these materials. One way to accommodate fibre variability would be to test the mechanical behaviour of samples from incoming batches of fibres and assign a grade to each batch, which could then be taken into account when the fibres are subsequently used to produce composite products. However, conventional methods of determining mechanical behaviour require test samples of constant cross-sectional area but, unfortunately, this is not the case for natural fibres which vary in shape, width and lumen size, from place to place along the fibre. Insight as to how to deal with such variability is provided from the textiles industry where strength is determined as a function of linear mass density rather than cross-sectional area. This paper examines the feasibility of using a similar approach for grading natural fibres for use in natural fibre composite products.