We read with great interest the article by Dungy et al1 in which they concluded that free samples of formula at hospital discharge do not have a negative effect on breast-feeding behavior. We have serious concerns about the analyses and interpretation of findings from this study. First, it is important that the authors report the rates of exclusive breast-feeding at hospital discharge. Figure 1 presents exclusive breast-feeding rates beginning at 2 weeks post partum.1(p727) Thus, it is impossible to interpret the results without knowing the baseline exclusive breast-feeding rates by group. Second, even though the authors fully acknowledge that their sample does not represent low-income minority groups, they indicate that the implication of their study is that it "brings into question the policy of banning infant formula discharge packages from hospital maternal fetal units" (abstract and page 729). It is unclear whether, by including the word "fetal" they are implying that free samples given out prenatally do not have a negative influence. If so, there are no data in their article to support this extension of their conclusion. Also, because previous studies2 have clearly identified low-income groups as being more vulnerable to free formula samples, their statement is unwarranted. Third, the authors seem to challenge the previous conclusions from a meta-analysis conducted by Pérez-Escamilla et al,2 which found that formula samples at discharge have a negative effect on any and full breast-feeding (not exclusive breast-feeding as the authors claim). They mention that 2 of the 5 studies included in the meta-analysis found no effects. However, the very reason that meta-analysis is such a powerful statistical technique for integrating results from individual studies testing the same hypotheses is that conclusions are not based on whether individual studies found significant results. Rather, the analysis treats all studies as a single unit from which a firm conclusion can be reached.3 Fourth, the authors imply in their results that providing the free formula sample may have even somehow promoted exclusive breastfeeding. However, Table 3 indicates that the group with the longest duration of exclusive breast-feeding was the one that received both a breast pump and the free sample and not the group receiving only the free sample of formula.1(p728) Furthermore, the between group differences in duration of exclusive breastfeeding were not statistically significant. For these reasons, the findings of Dungy et al need to be interpreted with caution.