Objective: The aim of this study is to describe a new objective method of detecting white matter injury by use of refiltering function of digital EEG. Study design: The neonatal EEGs of 9 patients with PVL were analyzed retrospectively, and compared with EEGs of 18 normal infants without PVL. Abnormal brushes were counted on the refiltered EEG where low cut filter were set at 10 Hz so that slow waves were eliminated. Abnormal brushes were defined as spindle-like fast wave bursts with maximal amplitudes higher than 40 mV and frequencies between 13 Hz and 20 Hz, which were easily distinguishable from a flat background. Results: The incidence of abnormal brushes in infants with PVL was 0.25/min, 0.45/min, 2.08/min, and 0.62/min on the frontal, central, occipital, and temporal region, respectively. The incidence in infants without PVL was 0.14/min, 0.08/min, 0.24/min, and 0.19/min, respectively. Abnormal brushes were observed more often in infants with PVL on the central, occipital, and temporal region than in infants without PVL. Localization of abnormal brushes was also correlated with the lesion of white matter injury on MR images and clinical outcome. Conclusion: Our results indicated that abnormal brushes on refiltered EEGs were strongly associated with white matter injury.