Objective: To examine the determinants of nutritional vulnerability in older refugees. Design: Cross-sectional data collection included anthropometry, functional ability measures (physical performance tests and independence in activities of daily living (ADL)), clinical screening, food frequency and socio-economic information. In-depth and group interviews were conducted in a sub-sample. To calculate body mass index (BMI) for the 5% with kyphosis, height was estimated from arm span using sex-specific regression equations from the non-kyphotic group. Setting: The study was carried out in the post-emergency phase in a Rwandan refugee camp in Karagwe district, north-west Tanzania. Subjects: 413 men and 415 women aged 50-92 years. Results: Results of this study may not be generalised as the refugees were most likely a self-selected group in that those with poorest nutritional status may have been left behind or have died on the way to the camp or soon after arrival and because of the presence of Help Age International. Physical test performance was lower in women than in men and lower in older age groups. Independence in ADLs was above 90% in both sexes. Functional ability problems mainly related to mobility. Individuals with poor nutritional status had an almost double risk of impaired handgrip compared to those with adequate nutritional status. The prevalence of undernutrition (BMI<18.5 kg/m2) was 19.5% in men and 13.1% in women and was higher above age 60 in both sexes (P<0.05 in men). Arm muscle area (AMA) was also significantly lower in older age groups. No difference was found in arm fat area. The proportion with low BMI was much higher in the group with kyphosis. Using multivariate techniques, handgrip strength proved to be the strongest independent determinant of nutritional vulnerability. People of poor nutritional status had less handgrip strength both absolutely as well as relative to their body size. Other significant determinants were food and health related and socio-economic factors. The total explained variance in men was 19.3% with BMI as nutritional status indicator and 26.4% with AMA, and in women 11.5% and 19.5% respectively. Older people themselves defined vulnerability according to the following main criteria: physical impairment, lack of purchasing power, and lack of support. A screening tool for entry into feeding or (social) support programmes is proposed based on anthropometric, clinical and social criteria. Conclusions: Undernutrition occurred among older refugees and was more prevalent at advanced age and among kyphotic people, illustrating the importance of including this group in nutritional status assessments. A relationship between poor nutritional status and impaired functional ability was demonstrated independent of age, sex and health conditions. Older refugees of poor nutritional status are likely to be more dependent, which will affect their quality of life.