SplitsTree is an interactive program for analyzing and visualizing evolutionary data. It implements the method of split decomposition, which allows for the representation of evolutionary data as a tree-like network when the data are not ideal. The program supports various analyses, including distance transformations, parsimony splits, spectral analysis, and bootstrapping. Two versions of SplitsTree are available: an interactive Macintosh version (shareware) and a command-line Unix version (public domain). Both are accessible via ftp and a web version. Split decomposition is a transformation-based approach that decomposes evolutionary data into a sum of 'weakly compatible splits' and represents them as a splits graph. For ideal data, this results in a tree, while less ideal data produce a tree-like network that can indicate conflicting phylogenies. SplitsTree version 2 is a new implementation based on the concepts of splits, splits graphs, and split decomposition. SplitsTree allows users to input sequences, distances, or split systems and outputs a weakly compatible system of splits and a splits graph. It includes various menus for data processing, distance transformations, and methods for generating split systems. The program also provides options for excluding certain taxa, sites, or constant sites, and for applying different distance transformations such as Hamming, Kimura 3ST, Jukes-Cantor, and LogDet. The program can be used to analyze biological data, revealing conflicting signals in evolutionary relationships. For example, in an rDNA dataset, SplitsTree revealed conflicting relationships between certain taxa, which were attributed to similar base compositions in their sequences. The program also includes a bootstrap feature to assess the statistical robustness of the splits graph. SplitsTree is based on the Nexus file format and can handle input data in the form of taxa, characters, or distances. It generates output files with specific blocks containing the split system, splits graph, and assumptions made during analysis. The program is implemented in C++ and is available for various Unix systems, with plans for a Windows version in the future. It has been developed through collaboration between researchers in Germany, New Zealand, and the University of Canterbury, with support from various funding bodies.SplitsTree is an interactive program for analyzing and visualizing evolutionary data. It implements the method of split decomposition, which allows for the representation of evolutionary data as a tree-like network when the data are not ideal. The program supports various analyses, including distance transformations, parsimony splits, spectral analysis, and bootstrapping. Two versions of SplitsTree are available: an interactive Macintosh version (shareware) and a command-line Unix version (public domain). Both are accessible via ftp and a web version. Split decomposition is a transformation-based approach that decomposes evolutionary data into a sum of 'weakly compatible splits' and represents them as a splits graph. For ideal data, this results in a tree, while less ideal data produce a tree-like network that can indicate conflicting phylogenies. SplitsTree version 2 is a new implementation based on the concepts of splits, splits graphs, and split decomposition. SplitsTree allows users to input sequences, distances, or split systems and outputs a weakly compatible system of splits and a splits graph. It includes various menus for data processing, distance transformations, and methods for generating split systems. The program also provides options for excluding certain taxa, sites, or constant sites, and for applying different distance transformations such as Hamming, Kimura 3ST, Jukes-Cantor, and LogDet. The program can be used to analyze biological data, revealing conflicting signals in evolutionary relationships. For example, in an rDNA dataset, SplitsTree revealed conflicting relationships between certain taxa, which were attributed to similar base compositions in their sequences. The program also includes a bootstrap feature to assess the statistical robustness of the splits graph. SplitsTree is based on the Nexus file format and can handle input data in the form of taxa, characters, or distances. It generates output files with specific blocks containing the split system, splits graph, and assumptions made during analysis. The program is implemented in C++ and is available for various Unix systems, with plans for a Windows version in the future. It has been developed through collaboration between researchers in Germany, New Zealand, and the University of Canterbury, with support from various funding bodies.