Chemical representations serve as a communication tool not only in exchanges between scientists but also in chemistry lessons. The goals of the present study were to measure the extent of student teachers’ knowledge about chemical representations, focusing on chemical formulae and structures in particular, and to explore which factors related to the education in school and university contribute to inter-individual differences. Using a quantitative cross-sectional design, 322 students from 12 German universities in different stages of their university education were tested with the Chemical Representations Inventory (CRI, Taskin, Bernholt, & Parchmann, 2015). In addition, a short questionnaire was administered containing demographic data and possible factors that could have an effect on students’ success in solving the items of the inventory. The data was analyzed by using Rasch modeling. The results show that student teachers’ knowledge about chemical representations is quite low, with an average of students’ total achievement of 50 % in the corresponding inventory on chemical representations. A multivariate linear model revealed that passing exam(s) in organic chemistry at university, the grade of school leaving certificate, gender as well as studying chemistry in upper secondary school on basic and advanced levels are significant predictors of student teachers’ knowledge. In total, these predictors are able to explain 30.3 % of the variance in the test results. The dominance of school-related variables in the regression analysis indicates that school education seems to be still important after several years of studying and is not equalized by education at university.