The proportion of very old people is rising, and so, describing their health-related quality of life (HRQoL) is an important point of interest. The aim of this study was to analyse the predictive factors on HRQoL throughout a 3-year follow-up period, in a community-based cohort of octogenarian people. From 290 subjects aged 85 and over, sociodemographic and geriatric data, including levels of frailty phenotype assessment, and HRQoL using the EuroQol 5D3L (EQ-5D) instrument were collected. A longitudinal analysis was performed by generalized estimating equations (jointly testing the bivariate effect of variables and its time dependence) and regression mixed models to evaluate the adjusted effect of variables on HRQoL after a 3-year follow-up. In the EQ-5D baseline assessment, the average visual analogue self-rating scale value was 63.82 (SD ± 19.45), the EQ-5D index was 0.67 (0.34) and the most significant issues were pain/discomfort (61.2 %), depression (45.3 %) and mobility (44.6 %). The third year index was 0.55 (0.38). Independent predictive factors of a lower HRQoL identified by the regression mixed models were female gender (marginal effect ME = -0.101; p = 0.003), being pre-frail (ME = -0.142; p = 0.011) or frail (ME = -0.071; p = 0.030), having heart failure (ME = -0.081; p = 0.037) and having a high social risk score (ME = -0.020; p = 0.015). In contrast, higher functional status (ME = 0.050; p < 0.001) and nutritional score (ME = 0.013; p = 0.011) appeared to be predictive factors of an enhanced HRQoL. The adjusted effect of "time of follow-up" had no statistical significance. Frail individuals at baseline have a significant lower HRQoL, whereas a higher functional status and nutritional status are independent predicting factors of an enhanced HRQoL after 3 years of follow-up. These findings may encourage clinicians in order to asses HRQoL.