The research aims to obtain information on the ability to understand the physics formulas of high school students being taught through direct learning models. The population subjects in this study were 360 high school students. The research sample was taken using a random class technique, with a total sample of 79 students. This research used instruments in the form of tests of understanding students' physics formulas in the form of essays that have been tested before being used in research to

more »

... etermine the validity and reliability of the tests. The normality test is performing on the experimental class which obtained Xcount2 < Xtable2 (1,68 < 9,49) and the control class Xcount2 < Xtable2 (1,47 < 9,49), the normality of the two data groups is normal. Homogeneity testing of variance used F-test get value 1,15, which Ftable = 1.74 (Fcount < Ftable), the variance of the two data groups is homogeneous. Data processing uses inferential analysis techniques with the "t" test. There is two class, which one class as an experimental class that was treated in the form of direct teaching-learning models and another class as a control class taught conventionally. The results there is an increase in the score of students' understanding of physics formulas taught through direct learning models. It can happen because students are invited to get accustomed to gaining procedural knowledge also can capture and interpret physics formulas that are taught through modeling or experiment.