This research is motivated by the lack of connection capabilities junior high school students in Bandung West. This study will examine the achievement of matemati connection capabilities of students usinglearning reciprocalteaching with the usual approach. The method used is an experimental method to design pretest-posttest control group, ie design involving at least two groups and randomly sampling classes. At the beginning and end of the study both classes were given the test. The population in this sample is the entire junior high school students in Bandung West whose characteristics have a low ability mathematical connections. Of the entire junior high school in West Bandung randomly selected and elected SMP Negeri Bandung Barat representing the characteristics of the population. The sample was selected randomly, where class G into the experimental class VIII, and VIII F be the class of the control. Collecting data in this study using a test description that has been tested first, and then score the mathematical connection capabilities of students were analyzed by descriptive and inferential statistics using two different test average. Based on the results, good data analysis and hypothesis testing, with α = 0.05 mathematical connection capability turns the achievement of students usinglearning model reciprocal teaching is better than the students who used the usual approach