It is well established that confounding factors related to head motion and physiological processes (e.g. cardiac and breathing activity) should be taken into consideration when analyzing and interpreting results in fMRI studies. However, even though recent studies aimed to evaluate the performance of different preprocessing pipelines there is still no consensus on the optimal strategy. This may be partly because the quality control (QC) metrics used to evaluate differences in performance across pipelines often yielded contradictory results. Importantly, noise correction techniques based on physiological recordings or expansions of tissue-based techniques such as aCompCor have not received enough attention. Here, to address the aforementioned issues, we evaluate the performance of a large range of pipelines by using previously proposed and novel quality control (QC) metrics. Specifically, we examine the effect of three commonly used practices: 1) Removal of nuisance regressors from fMRI data, 2) discarding motion-contaminated volumes (i.e., scrubbing) before regression, and 3) low-pass filtering the data and the nuisance regressors before their removal. To this end, we propose a framework that summarizes the scores from eight QC metrics to a reduced set of two QC metrics that reflect the signal-to-noise ratio (SNR) and the reduction in motion artifacts and biases in the preprocessed fMRI data. Using resting-state fMRI data from the Human Connectome Project, we show that the best data quality, is achieved when the global signal (GS) and about 17% of principal components from white matter (WM) are removed from the data. In addition, while scrubbing does not yield any further improvement, low-pass filtering at 0.20 Hz leads to a small improvement.In this work, we examined the performance of model-free and model-based techniques using QC metrics previously proposed and novel metrics related to large-scale network identifiability and presence of motion artifacts and biases. Multisession resting-state fMRI data from the Human Connectome Project were considered . With respect to model-free approaches, we examined FIX ("FMRIB's ICA-based X-noisefier"; Salimi-Khorshidi et al., 2014) as well as variants of aCompCor. FIX consists of whole-brain ICA decomposition followed by removal of noisy components identified using a multi-level classifier (Salimi-Khorshidi et al., 2014). Anatomical CompCor (aCompCor) refers to removal of the first five principal components from two noise regions of interest (ROIs), namely the WM and CSF compartments (Behzadi et al., 2007). Here, apart from evaluating the performance of the original aCompCor approach, we sought to answer whether removing more components would be beneficial examining components from WM and CSF separately. Finally, for the variant of aCompCor that exhibited the best improvement in QC scores, we investigated the additional benefit of removing nuisance regressors derived from the MPs and physiological recordings, excluding motion-contaminated volumes from the analysis and doi...