The thermodynamic interactions in blends of saturated hydrocarbon polymers originate from induced-dipole forces, and they differ in subtle but important ways, depending on the component structures. In this paper we use six model polyolefins, four statistical copolymers and two alternating copolymers, to examine some of the ways that have been suggested for organizing and interpretating these structural effects-statistical segment length mismatch, random copolymer theory, and the solubility parameter formalism. The polymers and their partially deuterated counterparts are components in an interconnected matrix of miscible binary blends for which the interactions were determined by small-angle neutron scattering. The interactions, quantified as Flory-Huggins interaction parameters, were broadly consistent with the length-mismatch idea, but some notable exceptions were found. Interpretations based on random copolymer theory provided homopolymer interaction parameters which conflicted with diblock copolymer results and other blend data. Tests of consistency with solubility parameter ideas were applied to the matrix of blends. The results for one component pair indicated strongly anomalous mixing, but those for all other pairs were consistent with the uniqueness criterion of the theory. Solubility parameters estimated from PVT measurements on the pure components agreed fairly well with assignments based on the measured interactions. Some inconsistencies remain, however, between diblock copolymer results and the solubility parameter formulation.