In this paper we report on the impact of a year-long teacher professional development (PD) workshop on student problem-solving skills. The PD workshop was held in Chile and promoted the use of collaborative problem-solving activities in the classroom, through monthly sessions during the school year. Participant teachers met with a monitor to solve a new problem collaboratively, plan a problem-solving activity for their next lesson, and reflect on their activities from their previous lesson. The study documents the performance of two groups of students when solving non-routine mathematics problems in two similar and non-equal tests consisting of three problems each, before and after the year of application of the workshop. The first group of students was the experimental group with collaborative problem-solving activities, as proposed in the PD workshop. The second group was the control group, composed of students whose teachers followed traditional teaching. In the first part of the paper we present the results of pre- and post-tests in which the students in the experimental group improved their problem-solving performance in a significant way compared with the control group. In the second part, we present a sample of representation strategy used by successful students in both groups and we discuss, in general, student strategies in solving test problems.