Epidemiological data on the association between fibrinogen levels and mortality are scarse and controversial. Longitudinal analyses were performed, separately by sex, on 17,689 individuals from the Moli-sani study [53% women, ≥35 years, free from cardiovascular disease (CVD) or cancer at enrolment], to evaluate the association between plasma fibrinogen and all-cause and cause-specific mortality. Over a median follow-up of 11.2 years, 1,058 deaths (34.7% CVD, 36.3% cancer) were ascertained. Both in the lowest (1.12-2.64 g/L) and highest (≥3.62 g/L) fibrinogen quintiles, women had an increased all-cause mortality hazard, when compared with third quintile (2.97-3.23 g/L). Dose-response analyses showed a U-shaped relationship in women (P overall <0.0001; P non-linear association <0.0001), but a positive linear association for all-cause mortality in men (P overall 0.0038; P non-linear association 0.76). Similar trends for a U-shaped association were observed for CVD mortality, while no association was observed with cancer deaths. A U-shaped association of fibrinogen levels with other-cause mortality was also found in both sexes. This study shows that not only higher but also lower fibrinogen levels represent hazard for mortality when compared to normal levels; U-shaped curves being prevalently observed in women.