This paper investigates the impact of head motion on resting-state functional connectivity (RSFC) in fMRI data. Motion-induced signal changes are complex, variable, and often persist for more than 10 seconds after motion stops. These changes increase RSFC correlations in a distance-dependent manner and are not effectively removed by motion-based regressors, but are reduced by global signal regression. The authors link data quality measures to motion, signal intensity changes, and RSFC correlations. They demonstrate that improvements in data quality measures may be cosmetic rather than true corrections. A within-subject, censoring-based artifact removal strategy reduces group differences due to motion to chance levels. The paper also notes conditions under which group-level regressions correct or do not correct motion-related effects. The study includes 160 healthy subjects across four cohorts: three adult cohorts (high, medium, low motion) and a child cohort. Data were collected using a Siemens 3T scanner with T1 and T2-weighted images for registration. RSFC BOLD runs were obtained with subjects fixating a crosshair. Functional images were processed with slice-time correction, realignment, intensity normalization, and resampling in atlas space. Further processing included demeaning, detrending, frequency filtering, and spatial blurring. Various regressors, including motion estimates, their derivatives, and tissue-based signals, were used to remove nuisance variables. The authors found that motion-related signal changes are widespread and persistent, affecting RSFC correlations. Motion-related effects are not fully removed by current regression strategies, and global signal regression is more effective. The study highlights the need for improved subject-level motion correction to reduce the need for group-level corrections. The paper also discusses the limitations of current data-driven methods for artifact reduction, such as matrix decomposition, signal variance removal, and volume censoring. The study emphasizes the importance of understanding motion-related variance and its impact on RSFC correlations. The results suggest that motion-related effects can persist for tens of seconds after motion stops and that current regression strategies may not fully correct these effects. The authors propose a censoring-based strategy to reduce motion-related group differences to chance levels. The study also notes that motion-related effects may be more pronounced in certain populations and that further research is needed to develop effective motion correction strategies.This paper investigates the impact of head motion on resting-state functional connectivity (RSFC) in fMRI data. Motion-induced signal changes are complex, variable, and often persist for more than 10 seconds after motion stops. These changes increase RSFC correlations in a distance-dependent manner and are not effectively removed by motion-based regressors, but are reduced by global signal regression. The authors link data quality measures to motion, signal intensity changes, and RSFC correlations. They demonstrate that improvements in data quality measures may be cosmetic rather than true corrections. A within-subject, censoring-based artifact removal strategy reduces group differences due to motion to chance levels. The paper also notes conditions under which group-level regressions correct or do not correct motion-related effects. The study includes 160 healthy subjects across four cohorts: three adult cohorts (high, medium, low motion) and a child cohort. Data were collected using a Siemens 3T scanner with T1 and T2-weighted images for registration. RSFC BOLD runs were obtained with subjects fixating a crosshair. Functional images were processed with slice-time correction, realignment, intensity normalization, and resampling in atlas space. Further processing included demeaning, detrending, frequency filtering, and spatial blurring. Various regressors, including motion estimates, their derivatives, and tissue-based signals, were used to remove nuisance variables. The authors found that motion-related signal changes are widespread and persistent, affecting RSFC correlations. Motion-related effects are not fully removed by current regression strategies, and global signal regression is more effective. The study highlights the need for improved subject-level motion correction to reduce the need for group-level corrections. The paper also discusses the limitations of current data-driven methods for artifact reduction, such as matrix decomposition, signal variance removal, and volume censoring. The study emphasizes the importance of understanding motion-related variance and its impact on RSFC correlations. The results suggest that motion-related effects can persist for tens of seconds after motion stops and that current regression strategies may not fully correct these effects. The authors propose a censoring-based strategy to reduce motion-related group differences to chance levels. The study also notes that motion-related effects may be more pronounced in certain populations and that further research is needed to develop effective motion correction strategies.