The thesis presents the problem of maintaining consistency in distributed database systems. It first compares new communications technologies with regard to distributed database consistency. Then it considers partitioning of communications network and provides the consequences such partitioning bears on processing of transactions in distributed database systems. Also, from the viewpoint of distributed database transactions, it presents ACID properties and attempts to further classify such transactions. The paper further analyses primary and derived horizontal fragmentation. Through an elaborate example the benefits and disadvantages of horizontal fragmentation have been described. An analysis of vertical fragmentation is presented as well. Next, query processing is analyzed. The concept of cost of a query is introduced. Next, a comparison between join relations is made, and related to it, a recommendation is made what technique would be preferred, given the circumstances. At that point, a comparison between static and dynamic query optimizers is presented. It is shown how different strategies impact speed and cost of processing of a query through a real-world example. Furthermore, query processing has been illustrated through Mariposa distributed database management system. Data replication in distributed database systems is an important topic. By replicating data across the sites, transactions have fast access to local copies. In this paper an analysis and comparison has been made between “eager” and “lazy” replication methods. “Eager” replica control provides data consistency in a straightforward way. Furthermore, via an “update everywhere” approach, all types of transactions can be submitted at any site without restrictions. Despite these characteristics, “eager update everywhere” replication is rarely used in commercial systems since existing solutions have severe disadvantages in terms of performance and complexity. Next, concurrency control algorithms are presented and compared: Two-Phase Lock (2PL) and Timestamps (BTO). Also, database deadlock is analyzed, together with the workable methods of its detection and resolution. Methods for preserving consistency of a distributed database are presented through Two-Phase Commit Protocol (2PC), which is analyzed for its strenghts and weaknesses. Similar analysis has been done for other consistency preserving protocols: PrC, PrA, PrAny, DPr2PC and 3PC, as well as for newest additions to the family - PROMPT protocol. Early Prepare (EP) protocol is presented and compared against PROMPT. PEP protocol, which has been developed as a combination of the two protocols (EP and PROMPT), is presented as well. At the end, recently developed protocol 1-2PC is analysed. This protocol is interesting due to its dual mode of operation – it can function either as one-phase or two-phase protocol.