Tight oil reservoirs need fracturing to obtain industrial productivity, and brittleness of rock has an important effect on fracturing. For oil reservoirs of the Permian Lucaogou Formation in Jimusar Sag of Junggar Basin, in order to predict brittleness index accurately, 19 typical tight oil core samples were selected and the related parameters of petrophysics and rock mechanics were measured at first, it is found that the static and dynamic brittleness indices vary greatly. Then, based on the static and dynamic experimental results of core samples and previous research results, the ratio of static and dynamic brittleness indices is constructed, it has a well correlation with porosity and clay content. Hence, according to the porosity and clay content correction, the static and dynamic conversion model of brittleness indices is built. The predicted results of the model are in good agreement with the experimental results. Then, on the basis of the composition, structure and deformation mechanism, the reservoirs are divided into three types via rock brittleness. The stress-strain curves of good, poor and moderate brittleness reservoirs are respectively linear, concave and "S". The static brittleness index-Poisson's ratio cross plot is built to classify the reservoirs. When the static brittleness index is greater than 85 and Poisson's ratio is less than 0.2, the reservoir shows good brittleness. When the static brittleness index is less than 40 and Poisson's ratio is greater than 0.24, the reservoir shows poor brittleness, and when the static brittleness index and Poisson's ratio are between them, the reservoir shows moderate brittleness. Finally, the static and dynamic brittleness index conversion model and reservoir classification standard are applied to formation evaluation in the study area, showing good application results. The research results are of guiding significance for the conversion of static and dynamic parameters of tight oil reservoirs, the selection of fracturing layers and fracturing operation schemes.