Objective: To investigate the impact of different levels of systolic blood pressure on all-cause, cardiovascular and cerebrovascular mortality in patients with nonvalvular atrial fibrillation (AF). Methods: This is a prospective cohort study. Patients with AF or atrial flutter diagnosed by 12 lead electrocardiogram during physical examination of Kailuan Group employees from July 2006 to December 2017 or previously diagnosed with AF in an inpatient setting at a level 2A hospital or above were eligible for the study. Baseline clinical characteristics including age, gender, systolic blood pressure were collected. According to the level of systolic blood pressure, patients were divided into systolic blood pressure<120 mmHg (1 mmHg=0.133 kPa)group, 120 mmHg ≤ systolic blood pressure<140 mmHg group, and systolic blood pressure ≥140 mmHg group. The time of first diagnosis with AF was defined as the start of follow-up and the final follow-up ended at December 2018. Primary endpoint was all-cause death. Related information was obtained through the social security system or inpatient medical records. The cause of death was defined according to the International Classification of Diseases disease (ICD-10) codes by professional medical stuffs. Multifactorial Cox proportional risk model was used to analyze the relative risk ratios for the occurrence of death in different systolic blood pressure level groups. The relationship between systolic blood pressure levels and mortality in the patients with AF was analyzed by using natural spline function curves. Results: A total of 1 721 patients with AF were enrolled (average age=(67.0±9.0) years), patients were followed up for (6.3±3.8) years. 544 out of 1 721 patients with AF died during the follow-up period (31.61%). The cumulative incidence rate of all-cause mortality, cardiovascular and cerebrovascular death was 26.13%, 25.59%, 36.96% and 14.86%, 11.87%, 19.76% respectively in the systolic blood pressure<120 mmHg, 120 mmHg ≤ systolic blood pressure<140 mmHg and systolic blood pressure ≥140 mmHg groups. The cumulative incidence rate of all-cause, cardiovascular and cerebrovascular death was significantly higher in the group with systolic blood pressure ≥140 mmHg than in 120 mmHg ≤ systolic blood pressure<140 mmHg group (P<0.05). Compared with 120 mmHg ≤ systolic blood pressure<140 mmHg group, multivariable Cox proportional hazards regression models showed that the HRs (95%CI) for all-cause, cardiovascular and cerebrovascular death were 1.47 (1.20 to 1.79) and 1.69 (1.27 to 2.26) for the group with systolic blood pressure ≥ 140 mmHg (P<0.05). In contrast, the HRs (95%CI) for all-cause, cardiovascular and cerebrovascular death in the systolic blood pressure<120 mmHg group were 0.99 (0.73-1.35) and 1.24 (0.82-1.89), respectively, with no statistically significant differences between the two groups (P>0.05). The natural spline curve showed that there was a "U" relationship between systolic blood pressure levels and all cause death and cardiovascular and cerebrovascular death in this patient cohort. Systolic blood pressure greater than or less than 123 mmHg was associated with increased risk of death of AF patients in this cohort. Conclusion: Compared with systolic blood pressure<120 mmHg and systolic blood pressure≥140 mmHg group, the risk of all-cause and cardiovascular and cerebrovascular death is the lowest in AF patients with 120 mmHg ≤ systolic blood pressure<140 mmHg in this cohort.