Background and purpose: Infarct volume and location have a weak relationship with motor deficit in patients with chronic stroke. Recent research has focused on the relationship between spared or seemingly "healthy" neural tissue and motor function. In this study we examined MRI scans of patients with chronic stroke to determine if characteristics of seemingly normal parenchyma could predict either response to different forms of upper extremity physical rehabilitation or to pre-treatment motor status. Methods: Individuals with chronic stroke (ages 60.6 ± 11.9 years) and mild/moderate upper extremity hemiparesis were administered either CI therapy (n = 14) or a comparison therapy (n = 29). The patients were assessed prior to and following therapy with MRI scans and the Wolf Motor Function Test (WMFT) Performance Time measure. Total voxels in combined grey matter (GM) and white matter (WM) segments (parenchymal volume) were divided by total voxels in GM, WM, and cerebrospinal fluid segments (intracranial volume) to obtain the brain parenchymal fraction (BPF). Results: BPF correlated with treatment gains on the WMFT (r(43) = -0.31, p = 0.04). Significant correlations between pre-treatment motor function and BPF were not observed. Conclusions: Individuals with greater BPFs after stroke show larger arm function gains after CI therapy, suggesting that reductions in volume of normal-appearing tissue may relate to ability to benefit from rehabilitation therapy in chronic stroke. © 2012 - IOS Press and the authors. All rights reserved.