To assess whether Roberts’ age-adjusted warfarin loading protocol is effective in Chinese patients and whether the SAMeTT2R2 score can predict international normalized ratio (INR) control. Roberts’ protocol for warfarin titration was applied to patients with non-valvular atrial fibrillation (NVAF) complicated with ischemic stroke at the Department of Neurology between 2014 and 2019. Clinical and sociodemographic variables were recorded. A minimum of 1-year follow-up was used to calculate the time in therapeutic range (TTR) of the INR. A total of 94 acute ischemic stroke patients with NVAF were included in the study. Seventy-seven (81.9%) of the patients had attained stable INR (2.0–3.0) at the fifth dose, and 90.0% of the patients had achieved stable INR on the ninth day. Seventeen (18.1%) of the patients had an INR > 4 during dose-adjustment period. Patients with INR > 4 had significantly lower body weight (53.8 vs. 63.1 kg, P = 0.014), lower rate of achievement of stable INR (35.3% vs. 92.2%, P = 0.000), and lower rate of TTR ≥ 65% (23.5% vs. 70.1%, P = 0.001), but with no significant increase in bleeding risk. A total of 89 patients underwent long-term INR follow-up, of which 58 (65.2%) patients achieved TTR ≥ 65%. Patients with poor TTR had significantly lower body weight (56.3 vs. 63.7 kg, P = 0.020) and lower rate of stable INR achievement (64.5% vs. 89.7%, P = 0.002). All 94 patients had SAMeTT2R2 score ≥ 2. There was no linear association between SAMeTT2R2 score and the rate of TTR ≥ 65% (Ptrend = 0.095). Chinese ischemic stroke patients with NVAF on warfarin can safely and quickly achieve therapeutic INR using Roberts’ age-adjusted protocol and can obtain a good TTR. Lower body weight may be a predictor of poor TTR and INR > 4. Patients who have not attained stable INR after adjusting the dose five times are at high risk for poor TTR. SAMeTT2R2 score may not predict TTR in Chinese ischemic stroke patients with NVAF.