The low results of student geometry learning require a trial of the think-pair-share method as a replacement for conventional methods. This research aims to: (1) compare the level of effectiveness of applying the think-pair-share method with conventional learning in improving student geometry learning outcomes, and; (2) determine whether or not there is a significant effect from using the two methods. The method used is quasi-experimental with a nonequivalent pretest-posttest control group design involving an experimental group and a control group. Each group consisted of 23 class VII students selected non-randomly from the population. Data was taken using a geometry learning outcomes test instrument. Data were analyzed using the N-Gain Score test, the results of the calculations were converted into interpretation categories for N-Gain effectiveness to be concluded. Research findings: (1) the application of both methods was "ineffective", as shown by the calculation results of the average N-Gain score (%) for the experimental group of 38.112% and the control group of 30.636%, and; (2) based on the results of the hypothesis test, it was concluded that there was no significant influence from the application of the two methods on improving student geometry learning outcomes. The implication of this finding is that it is not recommended to apply the think-pair-share method if it is only focused on improving the geometry learning outcomes of class VII students, but it can still be applied for the purpose of developing students' collaboration skills.