Abstract: It is very important to quantificationally characterize the relationship between complexity and predictability of macroscopic transportation system. The Lempel-Ziv algorithm of symbolic dynamics was introduced to measure the complexity of macroscopic transportation system. Aiming at the misapplications of this algorithm ,the general method of trial and error is put forward .The ARIMA model was applied to evaluate and predict the time series of macroscopic traffic volume. Complexity and prediction errors of 5 actual time series were calculated. Based on the comparison of the calculated results, it is assumed that complexity is negatively correlated with predictability in macroscopic transportation system.