The mathematics ability test instrument created by the teacher is without good and reliable instrument development mechanisms and procedures. This research aims to produce a valid and reliable test instrument to measure junior high school students' mathematical abilities. To make it, development research is carried out. The test instrument development model modifies the Wilson Model and the Oriondo and Antonio Model. There are three stages: initial development of the test instrument, testing of the test instrument, and wide-scale testing. The test sample was limited to 200 students from 4 (four) State Middle Schools in Lubuklinggau City. Meanwhile, the test is limited to 300 students from 6 (six) State Middle Schools in Lubuklinggau City. The trial result data was analyzed quantitatively. Item analysis uses a dichotomous scale. Data were analyzed using the R Program or Bilog MG Program and IBM SPSS Statistics 24. The data analysis technique consists of several aspects. These are test item suitability (goodness of fit), reliability, item characteristic curve, level of difficulty, and information function and SEM. The result is a test instrument for odd-semester middle school students' mathematical abilities. It can be concluded that the validity of the contents of the instrument items is based on the Aiken index which was assessed by 4 experts as valid. The construct validity of the instrument using Exploratory Factor Analysis is valid. The reliability coefficient of the odd semester middle school students' mathematical ability test instrument is 0.883, which means that the test instrument is reliable. The conclusion is that the test instrument produced is valid and reliable for measuring junior high school students' mathematical abilities.