OBJECTIVES: This study investigates the association between eating out and depressive symptoms among elderly people. Potential mediators that may link to elderly eating out and depressive symptoms are also discussed. METHODS: A unique dataset of 1,184 individuals aged 65 and older was drawn from the National Health and Nutrition Survey in 2008 in Taiwan. A bivariate probit model and an instrumental variable probit model were estimated to account for correlated, unmeasured factors that may be associated with both the decision and frequency of eating out and depressive symptoms in the elderly. An additional analysis is conducted to check whether the nutrient intakes and body weights can be seen as mediators that link the association between eating out and depressive symptoms of the elderly. RESULTS: Elderly people who eat out are 38 percent points more likely to have depressive symptoms than their counterparts who do not eat out, after controlling for socio-demographic characteristics and other factors. A positive association between the frequency of eating out and the likelihood of having depressive symptoms of the elderly is also found. It is evident that one additional meal away from home is associated with an increase of the likelihood of being depressed by 3.8 percentage points. With respect to the mediations, we find that nutrient intakes and body weight are likely to serve as mediators for the positive relationship between eating out and depressive symptoms in the elderly. CONCLUSION: Our results show that elderly who eat out have a higher chance of having depressive symptoms. To prevent depressive symptoms in the elderly, policy makers should be aware of the relationship among psychological status, physical health and nutritional health when assisting the elderly to better manage their food consumption away from home. LIMITATONS AND IMPLICATIONS FOR FUTURE RESEARCH: Our study have some caveats. First, the interpretation of our results on the causality issue calls for caution in that our analysis relies on a cross-sectional survey. Second, other measures to define elderly depression, such as the Center for Epidemiological Studies -Depression (CES-D) score, can be used to check the robustness of our findings. Finally, the availability of food outlets in the local area and family characteristics are possible associated with food away from home of the elderly. If data permit, the relationship between eating out and elderly depressive symptoms can be better identified after controlling for variables related to food facilities and family characteristics.