Selecting superior individuals requires a careful genetic evaluation. For this reason, repeated measurements of biometrical characteristics need to be carried out in the same individual. The repeatability of the measurements is represented in the repeatability coefficient. This index is a strategic tool to assists the accuracy of the observations and further selection. Our objectives were: to estimate the repeatability coefficient in many seed characteristics for black wattle clones; to test the best method to estimate the optimal number of measurements required; and to analyze genetic dissimilarity. Seed weight, longitudinal length, transversal length, thickness and hilum dimensions in 50 seeds of each one of nine clones were evaluated. The characteristics were evaluated using analysis of variance and of principal component method based on correlation and covariance matrix, and structural analysis. Mahalanobis distance and graphic dispersion were generated to evaluate genetic dissimilarity. We concluded that the most effective method to determine the repeatability coefficient was PCA based on covariance matrix. To achieve a 90% determination coefficient 20 measurements are needed. Based in our measures of genetic dissimilarity it was possible to differentiate clones among themselves and to recognize two genetically distinct groups.Selecting superior individuals requires a careful genetic evaluation. For this reason, repeated measurements of biometrical characteristics need to be carried out in the same individual. The repeatability of the measurements is represented in the repeatability coefficient. This index is a strategic tool to assists the accuracy of the observations and further selection. Our objectives were: to estimate the repeatability coefficient in many seed characteristics for black wattle clones; to test the best method to estimate the optimal number of measurements required; and to analyze genetic dissimilarity. Seed weight, longitudinal length, transversal length, thickness and hilum dimensions in 50 seeds of each one of nine clones were evaluated. The characteristics were evaluated using analysis of variance and of principal component method based on correlation and covariance matrix, and structural analysis. Mahalanobis distance and graphic dispersion were generated to evaluate genetic dissimilarity. We concluded that the most effective method to determine the repeatability coefficient was PCA based on covariance matrix. To achieve a 90% determination coefficient 20 measurements are needed. Based in our measures of genetic dissimilarity it was possible to differentiate clones among themselves and to recognize two genetically distinct groups.