Background: Standard selection criteria for thrombolysis typically exclude patients with acute ischemic stroke with unclear onset. Multimodal MRI screening may be able to identify those with a favorable benefit-risk ratio for thrombolysis. We aimed to evaluate the safety and efficacy of MRI-based thrombolysis in unclear-onset stroke (UnCLOS). Methods: We reviewed the thrombolysis database registries from 3 medical centers in Korea. Subjects received thrombolysis with intravenous tissue plasminogen activator (tPA) or combined intravenous tPA and intra-arterial urokinase within 3 h, or intra-arterial urokinase within 6 h from symptom detection. For patients with UnCLOS, MRI-specific eligibility criteria (i.e. positive perfusion-diffusion mismatch and absence of well-developed fluid-attenuated inversion recovery changes of acute diffusion lesions) were applied. Rates of immediate and 5-day recanalization, early neurological improvement and symptomatic intracranial hemorrhage (ICH) within 48 h after treatment and 3-month modified Rankin Scale (mRS) scores were compared between patients with UnCLOS and those with clear-onset stroke (CLOS). Results: 32 patients with UnCLOS and 223 patients with CLOS were included. Baseline characteristics were comparable between the two groups, except that the proportion of MRI screening was higher, and detection-to-door time and door-to-needle time were longer in the UnCLOS group (p < 0.01). Rates of recanalization (immediate, 81.3 vs. 63.1%; delayed, 80.6 vs. 69.1%), early neurological improvement (on day 1, 46.9 vs. 35.9%; on day 7, 50.0 vs. 49.3%), symptomatic ICH (6.3 vs. 5.8%) and 3-month outcome (mRS 0–1, 37.5 vs. 35.0%; mRS 0–2, 50.0 vs. 49.3%) did not differ between the UnCLOS and CLOS groups. Conclusion: These preliminary results suggest that thrombolysis based on MRI criteria may safely be applied to acute stroke patients with unclear onset.