Background People who are homeless and using substances frequently encounter barriers to accessing support. This paper aims to inform policy and practice by analysing changes in the tobacco, alcohol and illicit drug use of people experiencing homelessness. Methods Data derive from a qualitative longitudinal study (undertaken 2020/2021) and involving telephone interviews (n=310) conducted with 34 people accommodated in two London hotels provided as part of a UK policy response to COVID-19. The hotels offered various supports, including opioid replacement therapy, prescribed alcohol, licensed nicotine replacement therapy, and e-cigarettes. Participants’ substance use data were organised by Iterative Categorization and subjected to a content analysis to identify patterns and themes. Results At entry to the hotel, 5/34 participants (14.7%) had never used alcohol nor illicit drugs; 10/34 (29.4%) had only ever used alcohol (mostly without a problem); 11/34 (32.4%) had ever used illicit drugs but without a problem; and 8/34 (23.5%) had ever had a problem with illicit drugs. Sub-groups had different socio-demographic characteristics, particularly regarding being/not being a UK national, sex, and homelessness duration. Tobacco smoking was common across all sub-groups (18/34; 52.9%). Participants were often anxious about living with others who were using substances, and some worried about their own substance use. Substance use was changeable, with more decreases than increases. Changes related to intrapersonal (psychological), interpersonal (social) and structural (resource-based) factors. For example, decreases were precipitated by people feeling motivated to change, separation from others who used drugs, and receiving treatment or support. Conclusions Findings indicate that various interventions and accommodation models may benefit people who are homeless and using substances. An initiative that combined shelter and basic amenities, pharmacological treatment, psychosocial support, and space where substances were not available and other people using substances could be avoided resulted in an overall reduction in substance use amongst those accommodated.