Introductory physics classes attempt to guide students while they develop the mental structures and schema necessary to consistently apply the Newtonian concept of force to various situations. Science education research has supported the effectiveness of teaching methods, like Peer Instruction, that require the students to be active participants. This study compared the effectiveness of Peer Instruction when two different types of discussions, peer-led and teacher-led, were used to discuss concept questions in a high school physics class. The study addressed three research questions: (a) does Peer Instruction with peer-led discussions have a positive significant effect on learning (b) does Peer Instruction with teacher-led discussions have a positive significant effect on learning (c) is there a significant difference in learning gains between peer-led and teacher-led discussions. Normalized gain scores between the Force Concept Inventory (FCI) pretest and posttest and the normalized gain scores between concept questions and benchmark tests supported the hypotheses that there was a significant increase between pretest and posttest scores for both groups, but the repeated measures test found no significant difference between the two treatment groups. An important limitation to the study was that there was no significant correlation between the pretest and posttest scores, which limited the ability of the study to support the hypotheses that the specific discussion format caused the change between pretest and posttest scores. Ways to improve the research design to increase the correlation between pretest and posttest scores for future research is discussed.