The rural infrastructure plays an increasingly important role in economic growth. This paper extends the partial least-square (PLS) method into rural infrastructure analysis. Because the economic factors which influent rural infrastructure usually are multicollinearity, PLS can avoids this problem. PLS extracts variables one by one from few sample data. Under the control of modeling, it makes fully use of the useful information contained in the raw data. The experiments show that this method is feasible in analysis of the relationship between infrastructure and rural economic than OLS or ANN.