Sisvar is a statistical analysis system developed in 1994 and released in 1996. Initially developed in Pascal, it was compiled with Borland Turbo Pascal 3.0. The primary goals of Sisvar's development were to create a software for the statistical experimental course at the Federal University of Lavras, develop a genuinely Brazilian free software program, and provide efficient and reliable statistical analysis tools for the Brazilian scientific community. Sisvar has gained acceptance due to its reliability, accuracy, simplicity, robustness, and user interactivity. The system includes modules for descriptive statistics, hypothesis testing, interval estimation, normality tests, kernel density estimation, linear regression, model selection, and multiple comparisons. It is particularly noted for its ability to handle interactions and nested effects in linear models, apply the Scott-Knott test, and perform regression analyses on sliced crossover or hierarchical effects. However, it has limitations, such as only supporting balanced data for ANOVA and a maximum of 200 variables. A new version of Sisvar is under development in Java, aiming to enhance interactivity, solve compatibility issues, and expand analytical capabilities. The Java version will be multiplatform, supporting Windows, Solaris, Mac OS X, and Unix/Linux. It will also include multivariate statistical methods and computational techniques for non-central distributions. The source code will be available for contributions from the community, fostering further development and improvement.Sisvar is a statistical analysis system developed in 1994 and released in 1996. Initially developed in Pascal, it was compiled with Borland Turbo Pascal 3.0. The primary goals of Sisvar's development were to create a software for the statistical experimental course at the Federal University of Lavras, develop a genuinely Brazilian free software program, and provide efficient and reliable statistical analysis tools for the Brazilian scientific community. Sisvar has gained acceptance due to its reliability, accuracy, simplicity, robustness, and user interactivity. The system includes modules for descriptive statistics, hypothesis testing, interval estimation, normality tests, kernel density estimation, linear regression, model selection, and multiple comparisons. It is particularly noted for its ability to handle interactions and nested effects in linear models, apply the Scott-Knott test, and perform regression analyses on sliced crossover or hierarchical effects. However, it has limitations, such as only supporting balanced data for ANOVA and a maximum of 200 variables. A new version of Sisvar is under development in Java, aiming to enhance interactivity, solve compatibility issues, and expand analytical capabilities. The Java version will be multiplatform, supporting Windows, Solaris, Mac OS X, and Unix/Linux. It will also include multivariate statistical methods and computational techniques for non-central distributions. The source code will be available for contributions from the community, fostering further development and improvement.