Regular exercise and community engagement may slow the rate of function loss for people with dementia. However, the evidence is uncertain regarding the cost-effectiveness and social return on investment (SROI) of home exercise with community referral for people with dementia. This study aimed to compare the social value generated from the in-person PrAISED program delivered before March 2020 with a blended PrAISED program delivered after March 2020. SROI methodology compared in-person and blended delivery formats of a home exercise program. Stakeholders were identified, a logic model was developed, outcomes were evidenced and valued, costs were calculated, and SROI ratios were estimated. Five relevant and material outcomes were identified: 3 outcomes for patient participants (fear of falling, health-related quality of life, and social connection); 1 outcome for carer participants (carer strain), and 1 outcome for the National Health Service (NHS) (health service resource use). Data were collected at baseline and at 12-month follow-up. The in-person PrAISED program generated SROI ratios ranging from £0.58 to £2.33 for every £1 invested. In-person PrAISED patient participants gained social value from improved health-related quality of life, social connection, and less fear of falling. In-person PrAISED carer participants acquired social value from less carer strain. The NHS gained benefit from less health care service resource use. However, the blended PrAISED program generated lower SROI ratios ranging from a negative ratio to £0.08:£1. Compared with the blended program, the PrAISED in-person program generated higher SROI ratios for people with early dementia. An in-person PrAISED intervention with community referral is likely to provide better value for money than a blended one with limited community referral, despite the greater costs of the former. Trial registration: ISRCTN Registry ISRCTN15320670.