Objective To investigate the efficacy and prognostic risk factors of high flow nasal cannula (HFNC) in elderly patients with respiratory failure. Methods Clinical data of 172 elderly patients with respiratory failure admitted to 363 Hospital from April 2020 to August 2022 were retrospectively collected. The patients were divided into an observation group (n=86) and a control group (n=86) according to treatment method. The observation group (54 males, 32 females), mean 68.67±2.36 years old, received HFNC oxygen therapy. The control group (52 males, 34 females), mean 68.12±2.14 years old, received conventional oxygen therapy. According to the prognosis after HFNC treatment, the observation group was subdivided into a poor prognosis group (n=21) and a good prognosis group (n=65). The clinical effects of different treatment methods in the two groups were analyzed, and the risk factors affecting prognosis of elderly patients with respiratory failure treated by HFNC were analyzed by multivariate logistic regression. A line graph model was constructed, and the model was verified by receiver operator characteristic curve and cumulative gain graph. Results Repeated measures ANOVA was conducted on the oxygen therapy indicators of the two groups of patients. The results showed that pH, PaO2, PaCO2, and respiratory rate all have statistical significance in terms of time effect (F=423.25, P<0.001; F=326.25, P<0.001; F=128.79, P<0.001; F=323.16, P<0.001), inter-group effect (F=128.79, P<0.001; F=205.46, P<0.001; F=310.52, P<0.001; F=123.15, P<0.001), and interactive effect (F=111.06, P<0.001; F=198.76, P<0.001; F=134.28, P<0.001; F=112.47, P<0.001). This indicated that the impact of time on pH, PaO2, PaCO2, and respiratory rate differs depending on the treatment method. The scores of acute physiology and chronic health evaluation Ⅱ (APACHEⅡ), the level of brain natriuretic peptide (BNP), the heart rate before treatment, the posterior root of tongue fall, the initial HFNC flow and the duration of HFNC in the poor prognosis group were significantly higher than those in the good prognosis group. Initial PaO2/FiO2 was significantly lower than that in the good prognosis group (P<0.05). Multiple factor analysis showed that APACHEⅡ score>17 points, BNP level>150 ng/L before treatment, heart rate >105 times/min before treatment, posterior root of tongue drop, initial HFNC flow>55 L/min, initial PaO2/FiO2<150 mmHg were independent influencing factors for poor prognosis of elderly patients with respiratory failure treated by HFNC. The histogram model showed that the total score of the above 6 indicators is 284, corresponding to a probability of poor prognosis of 71.6%, which proved that the prediction ability of this model is good. Conclusions The application of HFNC in elderly patients with respiratory failure has a significant effect. APACHEⅡ score, BNP level before treatment, heart rate before treatment, posterior root of tongue fall, initial HFNC flow, initial PaO2/FiO2 are all risk factors affecting the prognosis, which should be paid attention to in clinic to improve the therapeutic effect.