Two reviews, one by Summerbell et al. and the other by Doak et al. came to very different conclusions about the effectiveness of childhood obesity interventions. The aim of this commentary is to assess the extent to which inclusion and exclusion criteria, and the definition of effective outcomes, explain discrepant results. Differences in results were compared by inclusion criteria and outcome definitions. The most important summary recommendations for inclusion/exclusion criteria were to exclude all non-peer review articles, to maintain a 6-month lower limit for duration of study, to include interventions from before 1990, to include pre-school age groups, to include pilot studies and to intervene in high-risk communities. Authors did not reach consensus regarding inclusion of aims not specific to preventing weight gain and the manner of assessment of anthropometric measures. Combining both reviews and applying agreed exclusion criteria leaves 30 interventions; 50% are positive. Excluding studies without an aim specific to preventing weight gain leaves 10/24 (42%) positive interventions. The differences in the results of these two reviews relate to the inclusion criteria and outcome assessments. These findings underscore the importance of the evidence considered in assessing interventions.