The problem of power flow composition analysis， or the analysis of the composition， source， motivation and fate of each branch or element in a power system， is the crucial issue of many problems in power system analysis. But owing to the special characteristics of electric power networks， so far no commonly accepted power flow composition analytic method has yet been reported. Four commonly used methods of network power flow composition investigation， namely， the DC distribution factor method， the current-based method， the AC sensitivity method and the power flow tracing method， are reviewed. Their properties， merits， applicability advantages and disadvantages are discussed from such perspectives as simplification and hypotheses adopted， the power balance method， allocation objects and digital results. Case studies are implemented to verify the results.