The beet webworm Loxostege stictialis (L． ) is one of the main migratory pests inNorth China． Outbreakpopulations of this pest can seriously damage host crops． On the basis of data on the first generation of beet webworm larva collected from 1977 to 2008，we constructed a matrix of migration probability by applying the Markov chain model and then used this to predict the population of the first generation in 2009－2011． By using the transition probability method of Markov chain theory，a transition probability matrix of 1 to 3 steps was constructed based on time series data on the abundance of the pests’first generation from 1979 to 1999． This was used to predict the beet webworm abundance inKangbao city from 2009 to 2011． All predictions were 100% accurate and the results show that the predictive accuracy ofthe data collected from 1980 to 2011was 89. 9%． The predicted population completely conformed to that actually observedin the field．