Background: Gemcitabine plus nab-paclitaxel (GN) demonstrated a significant improvement of OS over gemcitabine alone (G) in the MPACT trial for metastatic pancreatic cancer. However, since patients aged 75 years or older was less than 10% in the trial, data for the efficacy and safety of GN in that population have been limited. Methods: We prospectively enrolled patients who were 76 years or older if they had pathologically proven, clinically unresectable pancreatic cancer and no prior history of chemotherapy. Treatment regimen was based on physicians' choice. The primary endpoint was OS and secondary endpoints included PFS, objective response, and safety. Geriatric assessments including G8, IADL, CCI and Mini-COG were performed at the time of registration and 3 months after. To adjust confounding factors in the comparison of GN and G, propensity score-matched analyses were performed. Results: We enrolled 233 eligible patients from 55 centers in Japan between September 2018 and September 2019. GN, G, other therapy, and BSC were administered to 116, 72, 29 and 16 patients, respectively. With propensity scores adjusted for age, sex, clinical stage, ECOG PS, CA19-9, and neutrophil-lymphocyte ratio, 42 patients each in GN and G were compared. Patients characteristics in the matched pair were well-balanced; median age, 79/79 years; ECOG PS 0, 43/36%; stage IV, 69%/64%. With a median follow-up of 14.8 months and 50 observed deaths, OS in GN showed a longer tendency than that in G: 12.2 vs. 9.4 months in median with a hazard ratio (HR) of 0.65 (95% CI, 0.37-1.13; p= 0.120). PFS in GN was significantly longer than that in G: 9.2 vs. 3.7 months in median with a HR of 0.38 (95% CI, 0.23-0.64; p= 0.0002). The objective response rates were 29% and 11% (p= 0.082), and the disease control rates were 81% and 61% (p= 0.088), respectively. Decline of G8 scores at 3 months was observed in 35% and 28% of patients in GN and G, respectively (p= 0.610), and that in IADL scores was observed in 61% and 37%, respectively (p = 0.194). GN had higher incidence of adverse events than G though all were not statistically significant: grade 4 neutropenia (23%/8%); grade 3-4 malaise (13%/8%), fatigue (15 %/10%), peripheral sensory neuropathy (5%/0%). The incidence of the other adverse events was comparable between GN and G: grade 4 leucopenia (3%/5%) and thrombocytopenia (0%/3%); grade 3-4 nausea (3%/5%), vomit (3%/5%) and febrile neutropenia (0%/0%). Treatment discontinuation due to adverse events was more often in GN than G (39%/14%). Conclusions: Our data suggests GN is more efficacious than G even in patients aged 76 years or older, although GN have higher incidence of grade 3-4 adverse events and tends to decrease geriatric assessment scores. Clinical trial information: UMIN000034265.