Multivariate analysis of leaf radian measurements was used to investigate variation in leaf shape among 34 Asian species of the Uvaria group, a large palaeotropical group of climbing Annonaceae characterized by imbricate petals and stellate hairs. Raw data were normalized by conversion into 15 ratio characters and using the log10 transformation. All species surveyed showed a unique leaf-shape ‘bauplan’. The ratio character with the greatest discriminating power in both the Principal Components Analysis and Discriminant Analysis (DA) results was a measure of the shape of the leaf base. Ratio characters with the highest factor loadings for principal components 1 and 2 clearly separated the sampled taxa when plotted against one another and provided support for the retention of several taxa as distinct species or varieties. Classification of cases into taxa using DA yielded a correct classification rate of only 52% for the ratio-transformed data; however, division of taxa in the dataset into smaller subgroups defined by discrete morphological characters significantly increased the accuracy of case identification to between 67 and 100% of cases correctly classified, depending on the group. Case identification using DA on log10-transformed data was higher than for the ratio values in the entire dataset (61.7%) and the larger subgroups. However, the rate of correct case assignment was lower in the smaller groups than for the ratio data.