Changes in ERP (P100 and N400) and root mean square (RMS) were obtained during a silent reading task in 28 patients with chronic post-stroke aphasia in a randomized, double-blind, placebo-controlled trial of both memantine and constraint-induced aphasia therapy (CIAT). Participants received memantine/placebo alone (weeks 0–16), followed by drug treatment combined with CIAT (weeks 16–18), and then memantine/placebo alone (weeks 18–20). ERP/RMS values (week 16) decreased more in the memantine group than in the placebo group. During CIAT application (weeks 16–18), improvements in aphasia severity and ERP/RMS values were amplified by memantine and changes remained stable thereafter (weeks 18–20). Changes in ERP/RMS occurred in left and right hemispheres and correlated with gains in language performance. No changes in ERP/RMS were found in a healthy group in two separated evaluations. Our results show that aphasia recovery induced by both memantine alone and in combination with CIAT is indexed by bilateral cortical potentials.