There are two main contributions in the thesis . Firstly, we design a novel and efficient algorithm for mining calendar-based association rules which have multilevel time granularities in temporal databases. Unlike apriori-like approaches , our method scans the database twice at most. By avoiding multiple scans over the database , our method can reduce the database scanning time. Secondly, we use membership functions to construct fuzzy calendar patterns which represent asynchronous periods. With the use of fuzzy calendar patterns, we can discover fuzzy periodical association rules which are association rules occurring in asynchronous periods. Experimental results have shown that our method is more efficient than others, and we can find fuzzy periodical association rules satisfactorily.