Severe alcoholic hepatitis (SAH) is often a progressive disease with high mortality and limited steroid responsiveness. Management options of steroid nonresponsive SAH (day 7 Lille score > 0.45) are limited. We assessed the efficacy and safety of granulocyte colony‐stimulating factor (G‐CSF) in steroid nonresponders. A randomized, double‐blind, single‐center trial (NCT01820208) was conducted between March 2013 and June 2016 in patients with histologically proven SAH, nonresponsive to 40 mg/day of prednisolone were randomized to G‐CSF (12 doses, 300 μg each in 28 days) or placebo. Responders were continued with prednisolone. Of the 430 patients with SAH, 132 received steroid therapy. Of these, 33 (25%) were nonresponders and were randomized to G‐CSF or placebo (14 in each group after exclusions). The baseline characteristics of both groups were comparable. The 28‐day mortality was comparable between the groups (21.4%, G‐CSF; 28.6%, placebo; P = 0.69). At 90 days, in the G‐CSF but not in the placebo group, the Model for End‐Stage Liver Disease reduced from 24.6 ± 3.9 to 19.4 ± 3.7 (P = 0.002) and Maddrey’s discriminant function from 74.8 ± 22.8 to 57.4 ± 31 (P = 0.26). Infections were less common (28% versus 71%; P < 0.001) with lower 90‐day mortality (35.7% versus 71.4%; P = 0.04) in the G‐CSF than in the placebo group. On Cox regression analysis, receiving G‐CSF (hazard ratio, 0.37; SD, 0.14‐0.98; P = 0.04), and high baseline serum creatinine (hazard ratio, 4.12; SD, 1.7‐10.3; P = 0.002) predicted day‐90 outcomes in steroid nonresponsive SAH. Patients tolerated G‐CSF without any major adverse events. Conclusion: Approximately one‐quarter of patients with SAH do not respond to corticosteroid therapy. Administration of G‐CSF is safe and helps to reduce the disease severity and 90‐day mortality in these patients.