It is well known that the behavior of page reference in a database environment is different from that in a program envionment; hence, some page replacedment algorithms, have been designed to improve the throughput and response time of a darabase management system. However, aspecial page replacement algorithm influenced by a dis scheduling algorithm has not been studied yet. For this reason, two new disk scheduling algorithms, called MAD and V-MAD, are introduced. The performace of a database management system is evaluated by pairing a page replacement algorithm with a disk scheduling algorithm under a multi- transaction environment. The best pair is found under the random, sequential, and random/sequemtial access patterns. The simulation results show that pairing OLRU or ILRU page replacement algorithms with SSTF, MAD, or V-MAD produces th highest throughput in random or random/sequential ac cess. in the case of sequential access, pairing CLOCK or GCLOCK with SCAN or V-MAD is the best. When the disk scheduling algorithms are MAD or V-MAD, the system's response time is shorter than that when using other methods such as FCFS, SSTF,SCAN, C-SCAN, N-STEP, or V(R) with R=0.2.