OBJECTIVES: Acute kidney injury requiring renal replacement therapy is a major concern in ICUs. Initial renal replacement therapy modality, continuous renal replacement therapy or intermittent hemodialysis, may impact renal recovery. The aim of this study was to assess the influence of initial renal replacement therapy modality on renal recovery at hospital discharge. DESIGN: Retrospective cohort study of all ICU stays from January 1, 2010, to December 31, 2013, with a "renal replacement therapy for acute kidney injury" code using the French hospital discharge database. SETTING: Two hundred ninety-one ICUs in France. PATIENTS: A total of 1,031,120 stays: 58,635 with renal replacement therapy for acute kidney injury and 25,750 included in the main analysis. INTERVENTIONS: None. MEASUREMENTS MAIN RESULTS: PPatients alive at hospital discharge were grouped according to initial modality (continuous renal replacement therapy or intermittent hemodialysis) and included in the main analysis to identify predictors of renal recovery. Renal recovery was defined as greater than 3 days without renal replacement therapy before hospital discharge. The main analysis was a hierarchical logistic regression analysis including patient demographics, comorbidities, and severity variables, as well as center characteristics. Three sensitivity analyses were performed. Overall mortality was 56.1%, and overall renal recovery was 86.2%. Intermittent hemodialysis was associated with a lower likelihood of recovery at hospital discharge; odds ratio, 0.910 (95% CI, 0.834-0.992) p value equals to 0.0327. Results were consistent across all sensitivity analyses with odds/hazards ratios ranging from 0.883 to 0.958. CONCLUSIONS: In this large retrospective study, intermittent hemodialysis as an initial modality was associated with lower renal recovery at hospital discharge among patients with acute kidney injury, although the difference seems somewhat clinically limited.