Shape is a crucial physical property of agricultural products and hence is an important parameter for assessing the quality standard. In the present study, shape variations among 17 hazelnut cultivars grown in Turkey were revealed from their digital images using shape descriptors obtained from elliptic Fourier analysis (EFA), which is a shape-based methodology. Subsequently, principal component analysis (PCA) was performed to summarize the variations among the hazelnut cultivars. This was followed by linear discriminant analysis using the first four principal components, representing 93.9 % of the total variance, obtained from the PCA to discriminate the 17 hazelnut cultivars. Estimates of Hotelling's pairwise comparisons from the multivariate analysis based on the shape variables obtained from the EFA revealed ideal shape differences between the hazelnut cultivars. Hierarchical cluster analysis divided the cultivars into six clusters according to their shape characteristics. In addition, size (projected area, length, width, thickness, surface area, geometric mean diameter), shape (shape index, sphericity, roundness, and elongation), and gravimetric (mass and volume) features of the 17 common hazelnut cultivars were also determined via an image processing technique.An analysis of variance was performed to test the differences among these variables in a descriptive method. We found that EFA provided excellent discrimination between the hazelnut cultivars with respect to their shape features.