Aim: To investigate the impact of nutritional status on self-reported as well as performance-based indicators of physical function in a rural elderly population in Bangladesh. Methods: A cross-sectional study conducted in Matlab, Bangladesh, included 457 randomly selected community-living elderly persons aged 60 years or more (60-92 years; 69 AE 7 years). Mobility, activities of daily living (ADL), performance tests, handgrip strength, the Mini Nutritional Assessment (MNA) and a structured questionnaire were used to assess physical function, nutritional status, socioeconomic status and health status, respectively. Descriptive and linear hierarchical regression analyses were applied. Results: Seven per cent of the participants reported limitations in mobility, and 8% reported limitations in ADL. However, more than half of the participants had difficulties in performing one or more items in the performance tests. According to the MNA, 26% of the participants were undernourished and 62% were at risk of malnutrition. More undernourished participants than well-nourished participants reported limited mobility, impaired ADL and difficulties in the performance tests. A corresponding reduction in grip strength was observed in the undernourished group. Accordingly, higher MNA scores, indicating better nutritional status, were significantly associated with higher mobility index, higher ADL index, higher performance tests index, and higher scores in handgrip strength. These associations remained after adjusting for demographic, socioeconomic and health status differences. Conclusions: Good nutritional status is important for the physical function of elderly people, even after controlling for possible confounders. Performance tests indicated a higher degree of functional impairment than that observed by self-reported estimation.