The effects of various artificial nutrition methods on the long-term outcomes of elderly patients are still not well known. We aimed to compare the long-term survival of the elderly newly administered with parenteral nutrition (PN) or enteral nutrition. This multicenter, prospective, observational cohort study was conducted on 546 elderly patients who were administered artificial nutrition. The main outcome was the survival ratio at 180 and 360 days after initiation of 3 different nutrition methods and estimated mean survival time: PN, nasal tube feeding (EN_N), and percutaneous endoscopic gastrostomy (PEG) feeding (EN_G). The incidence of systemic infection was also compared among different cohorts. At 180 and 360 days after initiation of artificial nutrition, the mortality rates in the PN, EN_N, and EN_G cohorts were 52% and 63%, 32% and 41%, and 22% and 33%, respectively. Multivariate logistic regression analysis showed that, whereas PN nutrition had significant associations with a higher death rate at 180 and 360 days in all samples, there is no significant difference on the main outcome among the 3 cohorts with neurological diseases. A subgroup analysis with neurological diseases showed that the proportional hazard ratios of the PN and EN_N cohorts in comparison with the EN_G cohort were 1.13 (95% confidence interval [CI], 0.66-1.92) and 1.22 (95% CI, 0.82-1.81). There is no significant superiority of PEG feeding compared with nasal tube feeding or PN. Clinicians should consider the choice of nutrition support method, taking into consideration the limitation of the patient's interest.