School garden-based education programs have become an increasingly popular tool to improve children's nutrition, yet evaluations have found mixed results. This study analyzes three years of student surveys collected before and after one year of participation in the American Heart Association's Teaching Gardens program. Analysis was guided by the Social Cognitive Theory, and measured changes in determinants of healthy eating behavior: preferences for fruits and vegetables, gardening skills, food systems learning, and perceptions of self-efficacy and social norms regarding fruit and vegetable consumption. A total of 755 pre-test and 976 post-test responses were included in the analysis, as well as 173 pre-test and 146 post-test responses from two control schools that did not participate in any gardening activities. Frequencies and crosstabulations were used to analyze pre- and post-test data. Paired t-tests were also used to analyze differences between pre- and post-test when data were collected for the same student before and after Teaching Gardens participation. Paired t-test results indicated improvement in Preferences, Food Systems Learning, and Social Norms for both control and experimental groups and in Gardening Skills for the experimental group, although none of the changes over time were significant. At posttest, significant differences by gender were found in responses across all five indices, with girls answering generally more positively to questions than boys. Crosstab results also indicated significant differences by school minority concentration and socioeconomic status (SES) across all five indices at post-test, as well as a general correlation between minority concentration and SES. However, the effects of these environmental factors were mixed. The second article of this thesis investigates the effects of another environmental factor - the level of integration of the school garden program into the wider school environment - on students' reported knowledge of Gardening Skills. Eight schools and 142 matched pairs of students that participated in the Teaching Gardens program were analyzed. Adult responses from each of the schools were used to create an index of the program's Level of Integration, which was then compared with the students' reported changes in Gardening Skills using bivariate analysis and Paired Samples t-tests. Repeated Measures General Linear Model tests were then conducted to compare the model including the Level of Integration against the more traditional model of school garden evaluation, which focuses on school's minority concentration and SES. Results indicate that students at schools with well-integrated school garden programs gain greater Gardening Skills as a result of one year of participation in the program and confirms previous findings that students from lower SES areas experience greater gains in Gardening Skills than students from higher SES areas.