Problem posing is a promising teaching method for enhancing motivation and performance in mathematics and more specifically in mathematical modelling. Hence, the goals of our study were twofold: (1) to examine the effects of problem posing on modelling performance, self-efficacy, and task values in solving modelling problems, and (2) to analyze whether problem posing affects modelling performance via self-efficacy and task values. In a randomized control trial involving ninth- and tenth-grade students (N = 210), participants were assigned to either a problem-posing and problem-solving group or to one of two problem-solving groups. Students in the problem-posing and problem-solving group received a booklet with descriptions of real-world situations and were prompted to pose and subsequently solve their own problems. Students in the two problem-solving groups received the same real-world situations with given problems and were asked to solve the problems. Before solving the problems, students in all groups reported their self-efficacy and task values. Prompting students to pose their own problems positively enhanced students' self-efficacy and partially improved their task values in solving modelling problems. Further, problem posing indirectly affected modelling performance via self-efficacy but not task values. However, problem posing had no total effect on modelling performance. The findings for self-efficacy and task values are in line with expectancy-value theories, adding new insights to the field by highlighting the importance of motivational constructs in problem-posing approaches and instructions aimed at fostering mathematical modelling.