The R package relaimpo provides six metrics for assessing the relative importance of regressors in linear models, with two—1mg and pmvd—recommended for decomposing R². It also offers bootstrap confidence intervals. The package is illustrated using the swiss dataset, which contains 47 observations on six variables, including fertility and other socio-economic factors. The paper discusses various metrics, including simple ones like first, last, betasq, and pratt, and more complex ones like 1mg and pmvd. The 1mg metric averages sequential R² contributions over all regressor orderings, while pmvd uses data-dependent weights to ensure regressors with zero coefficients receive zero importance. The paper also covers bootstrapping for assessing variability in relative importance metrics and provides comparisons with other R packages. relaimpo is noted for its efficiency and ability to handle correlated regressors, offering bootstrap confidence intervals and improved computation times for 1mg compared to hier.part. The package is recommended for decomposing R² and assessing relative importance in linear models.The R package relaimpo provides six metrics for assessing the relative importance of regressors in linear models, with two—1mg and pmvd—recommended for decomposing R². It also offers bootstrap confidence intervals. The package is illustrated using the swiss dataset, which contains 47 observations on six variables, including fertility and other socio-economic factors. The paper discusses various metrics, including simple ones like first, last, betasq, and pratt, and more complex ones like 1mg and pmvd. The 1mg metric averages sequential R² contributions over all regressor orderings, while pmvd uses data-dependent weights to ensure regressors with zero coefficients receive zero importance. The paper also covers bootstrapping for assessing variability in relative importance metrics and provides comparisons with other R packages. relaimpo is noted for its efficiency and ability to handle correlated regressors, offering bootstrap confidence intervals and improved computation times for 1mg compared to hier.part. The package is recommended for decomposing R² and assessing relative importance in linear models.