ABSTRACT Leaf area measurements are used in agronomic studies to evaluate plant growth, light interception, photosynthetic rates, and plant transpiration. It constitute an important indicator of crop productivity, for which the evaluation method must be fast, accurate, and of low cost. The objective of this study was to compare different indirect methods to estimate leaf area in pornunça (Manihot sp.). The research was carried out under field conditions from August 2017 to January 2019 in the semiarid region of Pernambuco State, Northeastern Brazil. Three methods were tested: linear dimensions of leaf (length, width, and the number of lobes), digital image, and leaf scanned image, analyzing 150 healthy leaves from 120 plants of pornunça at different growth stages. The criteria used to determine the best model(s) were a high coefficient of determination, low Akaike information criterion, low sum of squares of residuals, and high Willmott index. Independent of the method of determination, the power models showed the best criteria of adequacy for estimating the leaf area of the pornunça. The digital image, using the power model (Y=LW0.77NL0.49, where L and W are the leaf length and width, and NL is the number of lobes in the leaf) was the best non-destructive method for estimating the leaf area in pornunça plants.