Abstract Introduction During the COVID-19 pandemic, electroencephalography (EEG) proved to be a useful tool to demonstrate brain involvement. Many studies reported non-reactive generalized slowing as the most frequent pattern and epileptiform activity in a minority of patients. Objective To investigate the prevalence of diffuse unreactive background attenuation or suppression and its correlation with outcome in a cohort of COVID-19 patients. Methods The EEGs recorded during the first year of the COVID-19 pandemic were retrospectively evaluated to identify the main pattern and focus on the occurrence of a low-voltage background, either attenuated (10–20 μV) or suppressed (< 10 μV). We sought a correlation between in-hospital mortality and low-voltage EEG. In a subsample of patients, biomarkers of inflammation, hypoxemia and organ failure were collected. Brain imaging was also evaluated. Results Among 98 EEG performed in 50 consecutive patients, diffuse unreactive slowing was the most prevalent pattern (54%), followed by unreactive attenuation or suppression pattern (26%), being the latter significantly correlated with an unfavourable outcome (p = 0.0004). Survivors showed significantly lower interleukine-6 values compared to non-survivors. Patients with attenuated EEG and non-survivors also showed lower PaO₂/FiO₂ values. Neuroradiological findings were very heterogeneous with a prevalence of lesions suggestive of a microangiopathic substrate. Conclusions EEG attenuation or suppression may be more frequent than previously reported and significantly associated with a poor outcome. SARS-CoV-2 infection may result in encephalopathy and reduced EEG voltage through mechanisms that are still unknown but deserve attention given its negative impact on prognosis.