In 1987, the International Union of Nutritional Sciences (IUNS) subcommittee on Nutrition and Ageing, in conjunction with the World Health Organization (WHO) global program for the elderly, embarked on the 'Food Habits in Later Life' (FHILL): a cross-cultural study to determine to what extent health, social and lifestyle variables, especially food intake, collectively predict survival amongst long-lived cultures. A total of 818 participants aged 70 years and over were recruited from five IUNS centres. Mortality data were collected after five to seven years. The cohorts included Swedes in Sweden (SWD), Greeks in Greece (GRS) and in Melbourne, Australia (GRM), Anglo-Celts in Australia (ACS) and Japanese in Japan (JPN). Information was obtained on health, lifestyle and diet at baseline. A Cox Proportional Hazard model containing ten potential predictors of survival, adjusted to age at enrollment and ethnicity/locality, was developed to analyse the survival data. Based on up to seven years survival data, it was found that being an elderly Greek in Australia conferred the lowest mortality risk and being an elderly Greek in Greece conferred the highest mortality risk. When the ten potential predictors of survival were entered into the Cox model, the memory score, the Mediterranean diet score, Activities of Daily Living (ADL) and general health status scores showed the greatest effects in significantly reducing mortality hazard ratios by 22%, 13%, 4% and 4%, respectively. For diet score, a one-unit change predicted a significant 13% difference in survival. Of the lifestyle (modifiable) variables entered in the multivariate model, exercise and social activity were not significant predictors of survival suggesting that diet is a more important predictor of survival than these variables. Another lifestyle variable, smoking, significantly increased mortality hazard ratios by 67%, making it a more important predictor of survival than diet. Being male (non-modifiable) also increased risk of death by 63%. Diet, particularly the Mediterranean Diet, operates irrespective and together with other factors as an appreciable contributor to survival, with a strength comparable to or greater than all other measured variables. The independence and strength of the predictiveness of food pattern for survival, and for this to be cross-cultural from Europe to Asia is a novel and important observation for food and health policy.