A multicomponent liquid mixture viscosity model based on the Eyring and nonrandom two-liquid (NRTL) theories was recently presented and evaluated over the entire concentration range using data from several polymer-solvent systems. In this paper, the original component-based Eyring-NRTL viscosity model is transformed to a segment-based Eyring-NRTL viscosity model for polymer-solvent and polymer-polymer systems. The segment-based approach provides a more physically realistic model for large molecules when diffusion and flow is viewed to occur by a sequence of individual segment jumps into vacancies rather than jumps of the entire large molecule. Evaluation of a segment-based Eyring-NRTL model with polyisobutylene-isooctane mixture viscosity data demonstrates an improvement in the correlation capability over the M-n range of 900-1200 000. Additional evaluations with polystyrene in styrene, poly(ethylene glycol) in oxolane (THF), poly(dimethylsiloxane)s in pentamer, and poly(dimethylsiloxane) blends provide further support for using the segment-based Eyring-NRTL viscosity model to correlate the viscosity of mixtures containing polymers. For several cases evaluated, the segment-based Eyring-NRTL viscosity model is also found to provide some predictive capability because the binary parameters are essentially independent of polymer Mn.