Objectives: The purpose of this pilot study was to investigate the feasibility and effects of computer-based cognitive rehabilitation (CBCR) in patients with symptoms of visuospatial neglect or homonymous hemianopia in the subacute phase following stroke. Method: A randomized, controlled, unblinded cross-over design was completed with early versus late CBCR including 7 patients in the early intervention group (EI) and 7 patients in the late intervention group (LI). EI received CBCR training immediately after inclusion (m = 19 days after stroke onset) for 3 weeks and LI waited for 3 weeks after inclusion before receiving CBCR training for 3 weeks (m = 44 days after stroke onset). Results: CBCR improved visuospatial symptoms after stroke significantly when administered early in the subacute phase after stroke. The same significant effect was not found when CBCR was administered later in the rehabilitation. The difference in the development of the EI and LI groups during the first 3 weeks was not significant, which could be due to a lack of statistical power. CBCR did not impact mental well-being negatively in any of the groups. In the LI group, the anticipation of CBCR seemed to have a positive impact of mental well-being. Conclusion: CBCR is feasible and has a positive effect on symptoms in patients with visuospatial symptoms in the subacute phase after stroke. The study was small and confirmation in larger samples with blinded outcome assessors is needed.