Abstract: To question that the association rules with time factor may be neglected in data mining to the traditional related rule,an improved temporal algorithm of association rules with time-constrain is proposed,which first judges if the modes obtained by Apriori algorithm has relatively high support degree in data set at a certain time region through calculating the average supported-degree of the modes,the average supported-degree of the data set and the concentration degree of the modes,etc.Secondly this algorithm goes on data mining in the data set in order to find more accurate mode correlated with time,thus gets association rules at single time zone or different time zone or acrossing time zone or affairs.Through experimental analysis,this algorithm is feasible,and has certain meanings in practical application.