Introduction. Simple, accurate and non-destructive models determining leaf area of plants are important in many experimental comparisons. Determining the individual leaf area (La) of A. deliciosa (A. Chev.) vines involves the measurements of leaf parameters such as leaf length (Ll) and width (Lw), or some combinations of these parameters. Materials and methods. A 2-year investigation was carried out in Italy during 2005 (calibration experiments) and 2006 (validation experiment) under open field conditions. It aimed at comparing existing predictive leaf area models for A. deliciosa leaves using non-destructive measurements, and assessing the accuracy of the optimum model selected using an independent dataset. Results and discussion. Regression analyses of (La) vs. (Ll) and (Lw) revealed several models that could be used for estimating the area of individual A. deliciosa leaves. A linear model with (Ll × Lw) as the independent variable [La = 0.82 (Ll × Lw) – 0.28] provided the most accurate estimate (R2 = 0.985, mean standard error = 25) of A. deliciosa leaf area. Validation of the model with (Ll) and (Lw) measured on leaves from other orchards grown in different locations showed that the correlation between calculated and measured areas was very high. Conclusions. With the model selected, agronomists and physiologists can estimate accurately and reliably the leaf area of A. deliciosa plants without the use of expensive instruments, e.g., a leaf area planimeter or digital camera with image measurement software.