Abstract: The empirical research on economic growth has always been a critical issue for economists.The convergence hypothesis of the neoclassical economic growth theory states that there is a negative correlation between GDP growth rate and ratio of initial capital labor.Thus the poor regions tend to grow faster than the rich ones.This empirical observation has been a very debatable economic issue among the economists.Some economists have come to the conclusion that there are both convergence and divergence according to empirical study conducted in various regions.First,this paper states and interprets the concept about σ-convergence,absolute β-convergence,conditional β-convergence and the interaction among of them.Second,through Barro and Sala-I-Martin convergence inspection model,this paper seeks to find whether the convergence forecasted by the neoclassical growth theory really exists across three provinces(Liaoning Province,Jilin Province and Heilongjiang Province) and their 36 cities in Northeast China,based on the data of the provinces from 1990 to 2009.The following conclusions are observed: 1) The σ-convergence existed in the periods of 1993-1996 and 1999-2004 with the divergence in the periods of 1990-1993,1996-1999 and 2004-2009 by turns.2) The absolute β-convergence did not exist in the period of 1990-2009,but the conditional β-convergence was proved to exist from 1990 to 2009 after introducing the rate of fixed asset investment,the net migration rate of population and the region virtual variable.The conditional β-convergence means that economic growth converges towards the steady state.It follows then that the economic disparities do not disappear rapidly across economic regions even though the conditional β-convergence exists.