Taking the internal and external spot month crude oil price as the variable, these variables were analyzed dynamically.  Whether there are cointegration relations between the variables were tested by maximum likelihood method of Johansen and Juselius (JJ).  The results show that dynamic equilibrium relations exist in them.  Through Granger causality test and building error correction model, it is proved that there are dynamic balance relations between the two variables.  The dynamic balance relations were decomposed by using impulse response function and prediction error decomposition to understand the interaction mechanisms and extent of the effect.