The Permian Jiamuhe Formation in Jinlong-2 Block is an unconventional volcanic reservoir that has been a major target of oil production in the northwest margin of the Junggar Basin, China. This study focuses on analyses of geological, well-logging and production data to understand the characteristics and to set up quantitative evaluation method on volcanic reservoirs. The geological data reveals that the lithology mainly includes andesitic (welded) breccia, andesitic/basaltic brecciated lava and andesite. The reservoir space is mainly composed of stomata and dissolution pore, followed by intercrystalline pore and intergranular pore, therefore is characterized by low to medium low porosity (ϕ) and ultra-low to low permeability (k). The mercury injection capillary pressure curves have shown that the microscopic pore and throat of effective reservoirs mainly express in fine pore type with unimodal coarse state, partial coarse state and bimodal partial coarse state. Moreover, limits (ϕ ≥ 8%, k ≥ 0.03 × 10-³μm²) of effective volcanic reservoirs have been determined by oil test and cross plot of porosity and resistivity. Thus, based on the analysis of macroscopic and microscopic characteristics, we optimize the K-means clustering analysis method and typical Fisher's discriminant method to set up quantitative classification and evaluation criteria and discriminant functions, which divide the volcanic effective reservoir into 3 classes accordingly. Our results have been applied to the development test area, the average daily oil production of deployed horizontal wells is 204 bbl/d, and the average cumulative oil production for one year is 68 Mbbl. This quantitative classification method may provide strong technical supports for the efficient development of volcanic reservoir.