The use of a monotonous and less attractive learning model results in low connectivity and low confidence. The purpose of this study was to determine the effect of the Think Talk Write (TTW) learnng model on the mathematical connection ability and self-confidence students class IV fraction material. The formulation of the problem in this study is wether there is an effect of the Think Talk Write learning model on the mathematical connection ability and self-confidence students class IV fraction material. This research is conducted to students class IV. This study is a quantitative research with Quasi Experimental Design in the form Nonequivalent Control Group Design. Data collection techniques method the test and questionnaries. Data analysis uses normality test, homogeneity test and hypothesis test. The results show that: 1) There is an effect of the Think Talk Write learning model on the mathematical connection ability, as evidenced by the results of the Independent Sample T-Test, there is a sig. (2-tailed) 0.038 <0.05 then Ho is rejected and Ha is accepted in the Paired Sample T-Test there is a sig. (2-tailed) 0.000 <0.05 then Ho is rejected and Ha is accepted. 2) There is an influence of the Think Talk Write learning model on self-confidence from the independent calculation there is a sig. (2-tailed) 0.000 < 0.05 means that Ho is rejected and Ha is accepted, while in Paired the results are Sig. (2-tailed) 0.000 <0.05 means that Ho is rejected and Ha is accepted. This research implies that the TTW learning model can improve the mathematical connection ability and self-confidence students class IV fraction material.