This paper explores the relationship between BMI and several health conditions among Union Army veterans who had medical examinations between 1891 and 1905. We find that BMI, when used as a proxy of nutrition, helps to explain morbidity and mortality differentials among veterans. There is evidence suggesting that among Union Army veterans extremely low or high BMIs were both associated with poor health, as indicated by a higher level of disability rating, higher risk of developing certain diseases, and higher mortality risk than those associated with having normal weight. Compared to veterans with normal weight, underweight veterans were more likely to be diagnosed with cardiovascular, respiratory and gastrointestinal diseases, but were less likely to be diagnosed with rheumatic and musculo-skeletal conditions at the first examination. High BMI levels are also associated with a higher risk of developing cardiovascular and rheumatic diseases, and higher mortality in the 20 years after the first examination. We performed a longitudinal analysis to study the association between earlier BMI as well as weight change and later development of diseases. The results suggest that, as a predictor of diseases, the explanatory power of BMI becomes lower the farther into the future we try to predict. Compared with those who maintained the same weight, veterans who gained weight were associated with a lower risk of being diagnosed with gastrointestinal diseases at their second examination.