BACKGROUND: The balance of benefits and harms associated with enteral tube feeding for people with severe dementia is not clear. An increasing number of guidelines highlight the lack of evidenced benefit and potential risks of enteral tube feeding. In some areas of the world, the use of enteral tube feeding is decreasing, and in other areas it is increasing. OBJECTIVES: To assess the effectiveness and safety of enteral tube feeding for people with severe dementia who develop problems with eating and swallowing or who have reduced food and fluid intake. SEARCH METHODS: We searched ALOIS, the Cochrane Dementia and Cognitive Improvement Group's register, MEDLINE, Embase, four other databases and two trials registers on 14 April 2021. SELECTION CRITERIA: We included randomised controlled trials (RCTs), or controlled non-randomised studies. Our population of interest was adults of any age with a diagnosis of primary degenerative dementia of any cause, with severe cognitive and functional impairment, and poor nutritional intake. Eligible studies evaluated the effectiveness and complications of enteral tube feeding via a nasogastric or gastrostomy tube, or via jejunal post-pyloric feeding, in comparison with standard care or enhanced standard care, such as an intervention to promote oral intake. Our primary outcomes were survival time, quality of life, and pressure ulcers. DATA COLLECTION AND ANALYSIS: Three review authors screened citations and two review authors assessed full texts of potentially eligible studies against inclusion criteria. One review author extracted data, which were then checked independently by a second review author. We used the 'Risk Of Bias In Non-randomised Studies of Interventions' (ROBINS-I) tool to assess the risk of bias in the included studies. Risk of confounding was assessed against a pre-agreed list of key potential confounding variables. Our primary outcomes were survival time, quality of life, and pressure ulcers. Results were not suitable for meta-analysis, so we presented them narratively. We presented results separately for studies of percutaneous endoscopic gastrostomy (PEG) feeding, nasogastric tube feeding and studies using mixed or unspecified enteral tube feeding methods. We used GRADE methods to assess the overall certainty of the evidence related to each outcome for each study. MAIN RESULTS: We found no eligible RCTs. We included fourteen controlled, non-randomised studies. All the included studies compared outcomes between groups of people who had been assigned to enteral tube feeding or oral feeding by prior decision of a healthcare professional. Some studies controlled for a range of confounding factors, but there were high or very high risks of bias due to confounding in all studies, and high or critical risks of selection bias in some studies. Four studies with 36,816 participants assessed the effect of PEG feeding on survival time. None found any evidence of effects on survival time (low-certainty evidence). Three of four studies using mixed or unspecified enteral tube feeding methods in 310 participants (227 enteral tube feeding, 83 no enteral tube feeding) found them to be associated with longer survival time. The fourth study (1386 participants: 135 enteral tube feeding, 1251 no enteral tube feeding) found no evidence of an effect. The certainty of this body of evidence is very low. One study of PEG feeding (4421 participants: 1585 PEG, 2836 no enteral tube feeding) found PEG feeding increased the risk of pressure ulcers (moderate-certainty evidence). Two of three studies reported an increase in the number of pressure ulcers in those receiving mixed or unspecified enteral tube feeding (234 participants: 88 enteral tube feeding, 146 no enteral tube feeding). The third study found no effect (very-low certainty evidence).  Two studies of nasogastric tube feeding did not report data on survival time or pressure ulcers. None of the included studies assessed quality of life. Only one study, using mixed methods of enteral tube feeding, reported on pain and comfort, finding no difference between groups. In the same study, a higher proportion of carers reported very heavy burden in the enteral tube feeding group compared to no enteral tube feeding. Two studies assessed the effect of nasogastric tube feeding on mortality (236 participants: 144 nasogastric group, 92 no enteral tube feeding). One study of 67 participants (14 nasogastric, 53 no enteral tube feeding) found nasogastric feeding was associated with increased mortality risk. The second study found no difference in mortality between groups. The certainty of this evidence is very low. Results on mortality for those using PEG or mixed methods of enteral tube feeding were mixed and the certainty of evidence was very low. There was some evidence from two studies for enteral tube feeding improving nutritional parameters, but this was very low-certainty evidence. Five studies reported a variety of harm-related outcomes with inconsistent results. The balance of evidence suggested increased risk of pneumonia with enteral tube feeding. None of the included studies assessed behavioural and psychological symptoms of dementia. AUTHORS' CONCLUSIONS: We found no evidence that tube feeding improves survival; improves quality of life; reduces pain; reduces mortality; decreases behavioural and psychological symptoms of dementia; leads to better nourishment; improves family or carer outcomes such as depression, anxiety, carer burden, or satisfaction with care; and no indication of harm. We found some evidence that there is a clinically significant risk of pressure ulcers from enteral tube feeding. Future research should focus on better reporting and matching of control and intervention groups, and clearly defined interventions, measuring all the outcomes referred to here.