Aim. To identify patterns of prosocial behaviours under collective quarantine conditions.Method. Survey data was collected from a sample of Italian adults during the March ̶ May 2020 COVID-19 lockdown in Italy. Participants reported on offline and online prosocial behaviours, Sense of Community Responsibility (SoC-R) and perceptions of community resilience. Latent class analysis (LCA) was used for data analysis.Results. A total of 4,045 participants completed the survey and 2,562 were eligible (72% female; mean age 38.7 years). LCA revealed four classes of prosocial behaviours: Money donors (7%), Online & offline helpers (59%), Online health information sharers (21%), and Neighbour helpers (13%). The classes were partially invariant across age groups (18 ̶ 35 and > 35 years). Being a man and higher SoC-R scores were associated with belonging to the Online & offline helper class. Members of this class also reported the greatest perceptions of community resilience.Conclusions. Results offer insight on the multidimensionality of prosociality under collective quarantine conditions. Online & offline helpers could be targeted for promoting sustained altruism and involvement in community organisations. For the other groups, programmes should aim to eliminate barriers to help others in multiple ways.