Abstract Background Living conditions in homeless shelters facilitate the transmission of COVID-19. Social determinants and pre-existing health conditions place homeless people at increased risk of severe disease. Described outbreaks in homeless shelters resulted in high proportions of infected residents and staff members. In addition to other infection prevention strategies, regular shelter-wide (universal) testing for COVID-19 may be valuable, depending on the level of community transmission and when resources permit. Methods This was a prospective feasibility cohort study to evaluate universal testing for COVID-19 at a homeless shelter with 106 beds in Berlin, Germany. Co-researchers were recruited from the shelter staff. A PCR analysis of saliva or self-collected nasal/oral swab was performed weekly over a period of 3 weeks in July 2020. Acceptability and implementation barriers were analyzed by process evaluation using mixed methods including evaluation sheets, focus group discussion and a structured questionnaire. Results Ninety-three out of 124 (75%) residents were approached to participate in the study. Fifty-one out of the 93 residents (54.8%) gave written informed consent; thus 41.1% (51 out of 124) of all residents were included in the study. Among these, high retention rates (88.9–93.6%) of a weekly respiratory specimen were reached, but repeated collection attempts, as well as assistance were required. Around 48 person-hours were necessary for the sample collection including the preparation of materials. A self-collected nasal/oral swab was considered easier and more hygienic to collect than a saliva specimen. No resident was tested positive by RT-PCR. Language barriers were the main reason for non-participation. Flexibility of sample collection schedules, the use of video and audio materials, and concise written information were the main recommendations of the co-researchers for future implementation. Conclusions Voluntary universal testing for COVID-19 is feasible in homeless shelters. Universal testing of high-risk facilities will require flexible approaches, considering the level of the community transmission, the available resources, and the local recommendations. Lack of human resources and laboratory capacity may be a major barrier for implementation of universal testing, requiring adapted approaches compared to standard individual testing. Assisted self-collection of specimens and barrier free communication may facilitate implementation in homeless shelters. Program planning must consider homeless people’s needs and life situation, and guarantee confidentiality and autonomy.