Current evidence, almost exclusively from observational studies, provides a rather mixed picture. From the few studies that have been able to look at fatal or non-fatal cardiovascular events, there is little indication that breast-feeding is associated with either an increased or decreased risk. With respect to blood pressure, the meta-analyses suggest a small but statistically significant lowering of around 1 mmHg SBP associated with having been breast-fed in infancy. However, there is a strong indication from the meta-analyses that even this small effect may partly be accounted for by publication bias. The strongest evidence for an effect of breast-feeding reviewed in this chapter is for serum lipids, where there is good evidence that being breast-fed is associated with an increase in serum total cholesterol in infancy. In childhood there appears to be no association, while in adults there is some indication of breast-feeding being associated with a small decline in total cholesterol levels. As already outlined at the start of the chapter, this whole area of research is made particularly difficult by the fact that breast-feeding can be defined in many different ways. Some studies use definitions that are equivalent to exclusive breast-feeding prior to weaning, while others define it as having ever been breast-fed. This problem of classification is likely to dilute any real associations that may exist. The other major problem is one of interpretation. A result implying that breast-feeding is a "good thing" for cardiovascular health could equally be construed as evidence for a "bad" effect of bottle-feeding. From these data alone, we cannot convincingly determine which conclusion is correct. This is not simply a philosophical debating point. As discussed above in relation to the interpretation of results from the randomised trial of infant feeding, the issue has implications for all research on this topic. Some progress in this area will be made if studies are conducted which define breast-feeding in a more precise and comparable way, and take account of the composition of alternative infant feeds. This will be most easily done by following up more recent study populations that were originally recruited to look at shorter-term effects of infant feeding on outcomes such as growth. With respect to randomised trial evidence, looking at the cardiovascular disease risk profiles of children (and later adults) who were part of the PROBIT trial in Belarus (see Chapters 5 and 10) is likely to prove fruitful.