In this study, liquid-vapor vertical upflow has been research with the intent of finding an improved method of modelling the interphase friction in two-phase vertical flow in nuclear thermal-hydraulic codes. An improved method of modelling interphase friction should allow for better prediction of pressure gradient, void fraction and the phasic velocities. Data has been acquired from several available published resources and analyzed to determine the interphase friction using a force balance between the liquid and vapor phases. Using the Buckingham Pi Theorem, a dimensionless interphase friction force was tested and refined before being compared against seven other dimensionless parameters. Three correlations have been developed that establish a dimensionless interphase friction force as a function of the Weber number, the Froude number and the mixture Froude number. Statistical analysis of the three correlations shows that the mixture Froude number correlation should be the most accurate correlation. The correlations have a weakness that makes them ineffective mostly for bubbly flow and some slug flow scenarios, while they should perform significantly better for annular flow cases. Comparisons have been made against the interphase friction calculations published in the manuals of RELAP5/MOD2, RELAP5/MOD3.3, RELAP5-3D and TRACE. The findings have generally shown that the equations in the manuals provide very inaccurate approximations of the interphase friction compared to the interphase friction that was found via force balance. When analyzing the source code of RELAP5/MOD3.3, several differences were noticed between the source code and manual, which have been discussed. Calculations with the source code equations reveal that the source code provides a modestly improved prediction of the interphase friction force, but still has significant errors. Despite the fact that the manual and source code equations indicate that RELAP5/MOD3.3 should perform poorly in modelling interphase friction, actual RELAP5/MOD3.3 model runs perform very well in predicting pressure gradient, void fraction, the liquid and vapor velocities and the interphase friction force. This is largely due to RELAP5/MOD3.3 being able to adjust parameters to converge to a solution that fits within the boundary conditions established in the input file. Modifications to the RELAP5/MOD3.3 code were first made with the three correlations developed using dimensionless parameters, and were tested with data points that the RELAP5/MOD3.3 flow regime map had predicted would be annular flow. While the mixture Froude number correlation has been analyzed to be the most statistically accurate of the three correlations, it was found that the Weber number correlation performed best when implemented into RELAP5/MOD3.3. In a parametric study of the Weber number correlation, it performed optimally at 150% of the original correlation, improving upon the original RELAP model in almost every metric examined. Additional investigations were performed with individual annular flow correlations that model specific physical parameters. Results with the annular flow physical models were inconclusive as no particular model provided a significant improvement over the original RELAP5/MOD3.3 model, and there was no clear indication that combining the models would provide significant improvement. / Ph. D.