The chemistry version of the Colorado Learning Attitudes about Science Survey (CLASS-Chem) is a new instrument designed to measure students' (novices') beliefs about chemistry and learning chemistry compared to those of experts (instructors). This survey is intended to measure the effects of students' beliefs on learning, and to understand how educational practices influence these beliefs in order to guide chemistry instruction. This instrument is a modification of a survey validated and extensively used in physics. The CLASS-Chem survey addresses a range of beliefs, including how chemists think about visualization, chemical reactivity, and molecular structure. The survey statements are validated using student interviews to ensure clarity in interpretation and response. Faculty testing of the statements is performed to ensure consistency in the "expert" response. Statements are grouped into statistically robust categories of student thinking. Preliminary use of the survey shows that most traditional chemistry teaching practices cause declines in student beliefs over the course of one semester and that students' self-reported major status correlates with their "Overall" and "Personal Interest" score on the survey.