Patients with advanced pancreatic adenocarcinoma have a poor prognosis and limited second-line treatment options. Evidence suggests a role for the Janus kinase (JAK)/signal transducer and activator of transcription pathway in the pathogenesis and clinical course of pancreatic cancer. In this double-blind, phase II study, patients with metastatic pancreatic cancer who had experienced treatment failure with gemcitabine were randomly assigned 1:1 to the JAK1/JAK2 inhibitor ruxolitinib (15 mg twice daily) plus capecitabine (1,000 mg/m(2) twice daily) or placebo plus capecitabine. The primary end point was overall survival (OS); secondary end points included progression-free survival, clinical benefit response, objective response rate, and safety. Prespecified subgroup analyses evaluated treatment heterogeneity and efficacy in patients with evidence of inflammation. In the intent-to-treat population (ruxolitinib, n = 64; placebo, n = 63), the hazard ratio was 0.79 (95% CI, 0.53 to 1.18; P = .25) for OS and was 0.75 (95% CI, 0.52 to 1.10; P = .14) for progression-free survival. In a prespecified subgroup analysis of patients with inflammation, defined by serum C-reactive protein levels greater than the study population median (ie, 13 mg/L), OS was significantly greater with ruxolitinib than with placebo (hazard ratio, 0.47; 95% CI, 0.26 to 0.85; P = .011). Prolonged survival in this subgroup was supported by post hoc analyses of OS that categorized patients by the modified Glasgow Prognostic Score, a systemic inflammation-based prognostic system. Grade 3 or greater adverse events were observed with similar frequency in the ruxolitinib (74.6%) and placebo (81.7%) groups. Grade 3 or greater anemia was more frequent with ruxolitinib (15.3%; placebo, 1.7%). Ruxolitinib plus capecitabine was generally well tolerated and may improve survival in patients with metastatic pancreatic cancer and evidence of systemic inflammation.