This study explores the role of time preferences in determining health behaviors including diet, physical activity and smoking by combining traditional health economics concepts with behavioral economics perspectives. While there is extensive literature that conclusively show that high discount rates are associated with addictive behaviors, the conclusions are mixed for health producing behaviors including diet and physical activity. Our study utilizes population level data from Mexico with time discounting estimated from answers to hypothetical money choices. The first study explores the relationship between time discounting and health behaviors by using Grossman’s model of health capital by employing an instrumental variable approach to address the endogeneity in their relationship. We find that people who have low discount rates are more likely to report eating more vegetables per week, more likely to exercise and have more exercise minutes and are more likely to quit smoking. These results indicate that interventions to lower discount rates via education or social safety nets might improve health behaviors. Our second study compares exponential, hyperbolic and quasi-hyperbolic functional forms of time discounting to investigate time-inconsistency in time discounting and its influence on health behaviors. While our data are found to be inadequate to fit a quasi-hyperbolic model, we find that in our data exponential form is strongly supported and hyperbolic form is weakly supported. We find also that people who have time-inconsistent hyperbolic preferences have worse health behaviors. Further, we find ‘Oportunidades’, a conditional cash transfer program in Mexico, helps people with hyperbolic time preferences in improving their diet related behaviors. Our last study explores whether people assess their health relative to that of their peers and how such evaluation influences their health behaviors. We find that comparison to peers leads to assessing health from a loss-gain frame. People who think they have worse health tend to forecast better health behaviors and people who think they have better health than their peers forecast sub optimal behaviors. These results indicate that interventions aimed at peer groups would be more effective if those solutions take into consideration the comparison asymmetry within those groups.