The current investigation supports the objectives of Project GEMS (Roberts, 2008), a grant funded program whose objectives include the development and validation of a protocol to identify students from underrepresented and diverse populations as gifted in the content areas of science, technology, engineering and mathematics. Identification of students from low-income and diverse populations as gifted has been a struggle with current assessment techniques (Baldwin, 2005). Project GEMS aims to address this problem through development of interest measures specific to the STEM areas for use within an identification protocol. The current project developed a measure to assess interest in mathematics. The construct of interest was targeted as it is correlated with many positive factors in education that lead to increased academic performance (Schunk, Pintrich, & Meece, 2008). Existing math interest inventories are designed for older populations, lack good psychometric properties and are atheoretical. To improve upon existing interest measures, Hidi and Renninger’s (2006) four-phase model of interest served as the theoretical basis to inform and guide the process of development and validation of a math interest inventory. A twenty-seven item self-report math interest measure was designed to assess the four phases of Hidi and Renninger’s interest model (emotion, value, knowledge, and engagement; 2006). Pilot and field testing of the measure were conducted in elementary schools selected on the basis of a high proportion of low-income students in a south central Kentucky school district. The sample consists of 1,429,429 students in grades two through six. The measure was hypothesized to evidence good internal consistency, a four-factor structure, and a significant and positive correlations between the Iowa Test of Basic Skills and the composite and subscales of the math interest inventory. The first hypothesis found support with an internal consistency reliability coefficient of .916 for the overall score. Results of confirmatory factor analysis supported a four-factor structure resembling Hidi and Renninger’s (2006) four phase model of interest and including the four components emotion, value, knowledge, and engagement. The correlations between the math scores from the Iowa Test of Basic Skills and the math interest inventory total score and scales partially supported the last hypothesis. The correlations were small and positive for the Values and Knowledge scales but small and negative for the Emotion and Engagement scales. The correlations for the total score of the math interest inventory were significant; however, their values had little practical significance. While the math interest measure evidences good reliability and support for the structure of the scales through confirmatory factor analysis, the current study did not provide evidence for a significant relationship with math achievement as measured by a standardized group administered math achievement test. These results are discussed in relation to limitations of the current study and recommendations for further investigation.