Abstract Background Chronic kidney disease (CKD) is prevalent worldwide, with patients facing significant mortality risk in intensive care units (ICUs). Early identification of high-risk CKD patients is crucial for improving clinical outcomes. The blood urea nitrogen to albumin ratio (BAR) is a simple and measurable indicator, but its relationship with 28-day mortality in CKD patients is not well established. This study aimed to investigate this association. Methods We conducted a retrospective analysis of eligible CKD patients from the MIMIC IV database. The association between the BAR and 28-day mortality was assessed using Kaplan-Meier survival curves, multivariable Cox regression models, and restricted cubic spline models. Results A total of 4,625 patients were included, with a 28-day mortality rate of 25.2%. Kaplan-Meier survival curve analysis indicated that patients in the high BAR tertile had significantly lower survival probabilities than those in the low BAR tertile. The adjusted Cox regression model showed that compared to low BAR patients (T1 ≤ 9.8 mg/g), those in T2 (10.0-17.4 mg/g) and T3 (≥ 17.5 mg/g) had increased risks of 28-day mortality, with HRs of 1.49 (95% CI: 1.26–1.76) and 2.04 (95% CI: 1.73–2.40), respectively. Restricted cubic spline analysis indicated a nonlinear association. Conclusion The BAR is significantly associated with 28-day mortality risk in ICU patients with CKD and may serve as a valuable tool for mortality risk stratification.