The paper discusses the use of region of interest (ROI) analysis with the SPM toolbox for functional brain imaging. Most functional imaging studies analyze effects across the entire brain, but when focusing on specific brain areas, voxel-by-voxel approaches have limited power due to multiple comparisons. Whole brain analyses often use image smoothing, but the optimal filter depends on the activation shape, which may not match standard kernels like Gaussian. ROI analysis directly addresses the question of whether a specific brain area is activated by defining a region and performing statistical tests on the mean time course of voxels within that region. This method increases statistical power by avoiding multiple comparisons and can be enhanced by weighting voxels based on expected activation shapes. The MarsBar toolbox, implemented within SPM99, allows users to define regions using previous SPM analyses, binary images, or simple shapes. It supports combining regions using algebraic operations and can extract time courses for further analysis. The toolbox also computes t or F statistics for multiple regions and plots results using the SPM interface. The authors hope that this toolbox will simplify ROI analysis, which is expected to become more important as hypotheses about activation locations become more specific, offering advantages in statistical power and data interpretation.The paper discusses the use of region of interest (ROI) analysis with the SPM toolbox for functional brain imaging. Most functional imaging studies analyze effects across the entire brain, but when focusing on specific brain areas, voxel-by-voxel approaches have limited power due to multiple comparisons. Whole brain analyses often use image smoothing, but the optimal filter depends on the activation shape, which may not match standard kernels like Gaussian. ROI analysis directly addresses the question of whether a specific brain area is activated by defining a region and performing statistical tests on the mean time course of voxels within that region. This method increases statistical power by avoiding multiple comparisons and can be enhanced by weighting voxels based on expected activation shapes. The MarsBar toolbox, implemented within SPM99, allows users to define regions using previous SPM analyses, binary images, or simple shapes. It supports combining regions using algebraic operations and can extract time courses for further analysis. The toolbox also computes t or F statistics for multiple regions and plots results using the SPM interface. The authors hope that this toolbox will simplify ROI analysis, which is expected to become more important as hypotheses about activation locations become more specific, offering advantages in statistical power and data interpretation.