Introduction: Nutrition plays an important role in the maintenance and improvement of human life expectancy. The 'Food Habits in Later Life' (FHILL) is a cross-cultural study conducted under the auspices of the International Union of Nutritional Sciences (IUNS) and the World Health Organization (WHO). Baseline data on food habits, health status and social variables were collected from five cohorts aged 70 and over (Japanese in Japan, Swedes in Sweden, Anglo-Celtic in Australia, Greeks in Australia and Greece). Objective: To identify protective dietary predictors amongst long-lived elderly people (n=785) from the FHILL population after controlling for ethnicity. Methods: The validated FFQ were used to collect data on food intakes in all cohorts except Japanese where the 3d weighed food record method was employed. Intakes in gram/week were calculated by multiplying the serving size by the weekly frequency of intake. These values were further translated into gram/day and were adjusted to 2500 kcal (10,460 kJ) for men and 2000 kcal (8,368 kJ) for women. Food items were grouped into nine food groups based on key features of the Traditional Mediterranean Diet (vegetables, legumes, fruits and nuts, cereals (including starchy roots), dairy products, meat, fish, monounsaturated: saturated ratio, and ethanol). All-cause mortality data were obtained from up to seven years follow-up. Alternative Cox Proportional Hazard model adjusted to age at enrolment (in 5-year interval), gender, and smoking was developed to analyse the survival data. Each Cox model was tested against controlling for cohorts' location and ethnicity. Results: Only for legumes intake was the result plausible, consistent and statistically significant across collective FHILL cohort's data. There is a 7% - 8% reduction in mortality hazard ratio for every 20g increase in daily legume intake with adjustment for location/ethnicity (RR 0.92; 95% CI 0.85 - 0.99) and without adjustment for location/ethnicity (RR 0.93; 95% CI 0.87 - 0.99). Conclusions: This longitudinal study shows that a higher legume intake is the most protective dietary predictor of survival amongst the elderly, regardless of their ethnicity. The significance of legumes persisted even after controlling for age at enrolment (in 5-year interval), gender, and smoking. Legumes have been associated with long-lived food cultures such as the Japanese (soy, tofu, natto, miso), the Swedes (brown beans, peas), and the Mediterranean people (lentils, chickpeas, white beans).