Abstract: More than 40 years of anthor's exploration practice reveals that correct evaluation and prediction must be based on the combination of positive and reverse thinking. Simple positire thinking, which stresses to look for gas according to hydrocarbon sources and neglects natural gas activities, once resulted in slow reacting and faults on shallow hydrocarbon. From tortuous exploration experience and deep introspection of the west Sichuan Depression, the unique regularities of gas pools in ultra dense areas are mastered step by step, and the paleozoic Tahe oilfield and the C P ordos deep basin gas are predicted accurately during the evaluation of oil and gas areas in China. It is essential to make hydrocarbon prediction on the basis of ample data and using conventional positive thinking. The deeper the thinking is, the strorger the evaluation and prediction function are. Line type positive thinking often runs into a stone wall, because the collected data for deep hydrocarbon are still odd scraps and discontinuous. If they are linked up by positive thinking, they may be images from some angle even false appearances. And if using analysis, hypothesis, disproof and simulation carefully or using the correlation to succeed models as well as often using the way of reverse thinking, they can be even more approaching to reality.