Enter a distance matrix and click "Cluster". Lines should be split on whitespace, with the first line containing column headers. The script should cope if the first column header is missing. Each row should start with a name that matches the column headers. Currently, incomplete matrices are not supported (e.g. must have both sides of the diagonal) and symmetry will be forced.
A walkthrough of the UPGMA method can be found here.
Citation: If used in any publication, please cite this URL (http://bioware.soton.ac.uk/upgma.html).