In this paper, we present an efficient pricing strategy, the pivot and probe Algorithm for Network Flow Problems(PAPANET), specifically for solving capacitated, linear network flow problem (NPs). The PAPANET begins with an initial relaxed network problem(RNP), consisting of all the nodes and initial candidate arcs(possibly a few least cost arcs form the original problem and a set of all the artificial and slack arcs). After an initial solution to the RNP is derived by pivoting, the PROBE procedure identifies a set of most violated arcs from the noncandidate arcs that are not considered to be in the current RNP, and adds them to the RNP. The procedure also discards a set of least favorable, zero flow, nonbasic arcs from the RNP. The new RNP is solved to optimality and the procedure continues until all of the dual constraints of the noncandidate arcs are satisfied by the dual solution to the RNP. The PAPANET effectively reduces the problem size, time per pivot, and solution CPU time by eliminating noncandidate arcs. Computational tests on randomly generated problems indicate that PAPANET achieves and average savings of 50-80% of the solution CPU time of that of a comparable standard network simplex implementation.