In this large sample of fit males, healthy lifestyles, including exercise and milk consumption, are critical for optimal bone density and bone size. There was also an apparent synergistic effect between milk consumption and exercise with regard to bone density at multiple skeletal sites as well as on the cortical thickness of the tibia. It was also found that smoking and excessive caffeine consumption could have a negative impact on the skeleton in this cohort.
Racial differences in bone mass, bone geometry and size still exist even when controlling for height and weight, although as shown in other studies, [28, 29] the magnitude of those differences is smaller after controlling for body size. The differences in tibial cortical thickness and circumference have not been shown previously and may help explain the lower fracture rates in African Americans. Our study confirmed the findings of other studies that racial differences in BMD exist, [27, 30] and there are also differences in bone size and geometry even after controlling for height and weight. Limited power precluded us from have race specific models for each skeletal site.
Weight and height were consistent predictors of bone density at all sites as has been shown in prior studies; however weight and height are also related to tibial size. This is not surprising since genetic factors, including height and weight, can account for between 60–80% of the variation in BMD. [13, 14, 31–33] In this study, significant racial differences in BMD at the total hip and tibial mineral content persisted even after controlling for weight.
The positive effect of exercise on BMD is supported by this study. [34–36] Even in a fit population, with BMD at the upper end of the normal range, higher levels of exercise were associated with higher bone mass. Periosteal circumference, or size of the tibia, as well as the tibial mineral content were significantly influenced by prior exercise levels presumably due to the impact of increased mechanical loading from exercise. Since there may be few periods in life where bone size can be altered, taking advantage of this relationship in young adults may be important to a future reduction in fracture risk.
In this cohort of males, milk was significantly related to cortical thickness, hip and heel BMD and tibial mineral content. These findings support prior studies suggesting that milk intake is an important factor in increasing bone density and bone size in young adults. [37, 38] Milk intake may be a marker for calcium intake alone or for an overall healthy diet, or may reflect the influence of some other important nutrient in milk such as protein or vitamin D. It is likely that the skeletal benefit of milk based calcium is a result of many factors. Although there was not a significant interaction, the negative effect of caffeine was minimized at higher levels of milk intake, as has been previously reported.  This may indicate that the negative effect of caffeine on bone may be due primarily to reduced milk intake or to a reduced absorption of calcium at higher levels of caffeine intake. The negative dose response relationship between caffeine and calcaneal BMD has been reported in other populations. [40, 41]
Smoking also had a deleterious effect on hip and spine BMD in these otherwise healthy young men, as has been previously reported. [18, 20, 40, 42] This may relate to reduced osteoblast function in males who smoke  or perhaps a more rapid metabolism of hormones (estrogen and testosterone) that may affect BMD in men. 
A unique aspect of our study was the ability to determine factors that predict tibial size and mineral content in these young men. A skeleton that is more resistant to fractures should not only have a higher mineral content or density but also a larger bone size (circumference) and cortical thickness. The critical periods during which bone size can be maximized are likely to occur at an early age, prior to attainment of peak bone mass and this could be evaluated in this young adult cohort. Tibial size (periosteal circumference) was found to be related to genetically modulated characteristics including height and weight, as well as lifestyle factors including exercise and nutrition.
Mathematical models, including linear regression, are limited to evaluating the determinants of bone health in an additive way. This may not represent the true biologic relationship. Therefore we examined a number of cross product terms. The milk -exercise cross product proved to be significant in the tibial mineral content and cortical thickness model as well as for BMD of the calcaneus, hip, and spine. The milk – exercise interaction has been reported elsewhere and is biologically plausible. [45–47] The mechanical stimulus created by the exercise forces the bone to remodel to adapt to these new loads and repair damage incurred by fatigue created by the strains of exercise. The higher milk consumption creates a greater supply of calcium and possibly other needed nutrients and therefore, a larger more dense bone. The significant cross product term of milk and exercise is indicative of a synergistic interaction between the daily milk consumption and regular exercise.
This study has a number of limitations. First, it is not possible, as with all cross sectional studies, to establish a temporal relationship between diet, exercise and the acquisition of greater bone mass. Second, our baseline questionnaire was self-administered and therefore is limited by individual recall. The dietary questionnaire was limited because of time constraints and did not allow an assessment of factors that may be important including vitamin D, protein, other nutrients and total caloric intake. In addition we only assessed exercise history and milk consumption in the past year and we cannot know whether these behaviors were consistent over time. An additional limitation of the study may be its external validity: our study population is clearly more fit than the normal college age population. Therefore these findings may not be readily generalizable, although this study indicates that lifestyle variables, including exercise, may still have an impact on BMD. Finally, the statistical power, calculated post hoc ranged from 57 to 92 %. Some of the risk factors including smoking and alcohol intake could not be fully explored because of relatively small sample sizes, limited variability and limited power.
In conclusion, BMD and bone size are determined by a complex combination of genetic, lifestyle, and nutritional factors. In our study height, weight, prior exercise and milk intake, smoking and caffeine intake were the most frequent and important predictors of bone mineral density and bone size in these fit young men. In addition the interaction between milk consumption and exercise history was a significant predictor of bone size at the tibia and BMD at all skeletal sites. This information can assist in confirming intervention strategies for parents, schools and pediatricians during the critical years of bone development and the attainment of peak bone mass as suggested by the Surgeon General's report on Bone Health . These bone healthy behaviors should include higher milk intakes, adequate levels of exercise, limited caffeine intake and avoidance of tobacco products. Promotion of these behaviors associated with higher peak bone mass in males may help prevent stress fractures and osteoporosis and related fractures later in life.