The aim of the study was to examine mathematics teacher candidates’ strategies used in solving non-routine problems. Descriptive survey method was used for examining strategies. The research was carried out with 104 mathematics teacher candidates studying at a state university in Turkey. Data were collected through Problem Solving Test which was consisted 10 non-routine problems. Each problem is associated with a different strategy. Problems were adopted from Posamantier& Krulik (1998). Mathematics teacher candidates were asked to solve problems using appropriate problem solving strategies and to indicate the strategies they use in problem solving process. Descriptive statistics were used to determine mathematics teacher candidates solution and strategies in problem solving. Solutions of problems were rated and categorized based on content analysis. Problem solving process were analyzed by two researchers independently. Then ratings were compared. First, solutions were classified as true or wrong. Second true or wrong solutions classified as “solving using true strategy” and “solving using wrong strategy”. After that “solving using true strategy” were categorized “indicated true strategy” and “indicated wrong strategy”. The research of this study indicated that mathematics teacher candidates have adequate problem solving skills. Some of teacher candidates can not solve problems correctly because they use the wrong strategy. In other words mathematics teacher candidates have limited abilities at the stage of “understanding of a problem” and “planning”. It was a due of they could not understand what the problem says and they could not choose an appropriate strategy for solving problem. Some of teacher candidates used right strategies in solving problems, even though they could not reach the right answer. It was an evidence that teacher candidates have lack of knowledge about “applying plan”. On the other hand some of teacher candidates could solve problems correctly but they could not indicate true strategy.