We evaluated the current used allometric models and proposed a reliable and accurate model using non-destructive measurements of leaf length (L) and/or width (W) for estimating the leaf area of the Jatropha plant (Jatropha curcas L.). For model construction, a total of 1200 leaves were randomly selected from different levels of the tree canopies and encompassed the full spectrum of measurable leaf sizes (0.19-367 cm(2)). Power models better fit the Jatropha leaf area (LA) compared to linear models; however, when the LW was used, the intercept was removed from the linear model to precisely estimate LA. To validate these models, two independent data sets of 300 leaves were used. We demonstrate that the currently used models are biased and underestimate the area of Jatropha leaves. We developed an unbiased, single power model (Y = x(beta 1)) based on two leaf dimensions [LA = (LW)(0.9660)] with high precision and accuracy as well as a random dispersion pattern of the residuals. The model may be simplified by using only one leaf dimension (Y = x(beta 1)), LA = L-1.9644 or LA = W-1.8929. However, when either W or L alone was used as the single leaf dimension, the power model predicted the LA with good accuracy at the expense heteroscedastic residual dispersion behavior. Published by Elsevier Ltd.