Mixed variables data occurs frequently in most of the fields including medical, industrial, agriculture, marketing and remote sensing. In traditional clustering algorithms, Euclidean distance function is used to compute similarity between two data elements. It gives better and reliable results if numeric attributes are present in the data set. Although, when attributes are non-numeric i.e., categorical or mixed, this measure is not suitable for capturing the similarity of data elements. This study, includes five distance measures viz. Gower, Podani, Huang, Ahmad and Harikumar which is used for mixed variables data. Further, they have been fused into the agglomerative hierarchical clustering algorithms. The performance of these distance measures has been compared by cluster validation methods: Cophenetic correlation coefficient, Connectivity, Average Silhouette Width and Dunn index. The empirical analysis has been carried on pearl millet data with quantitative and qualitative variables using R software package. The experimental results showed that using three cluster validation measures; Cophenetic correlation coefficient, Connectivity, and Average Silhouette width, the Ahmad distance performed better than the Gower, Huang, Ahmad, and Harikumar distances.