The feasibility of evaluating thermodynamic interactions between polymer species by small-angle neutron scattering (SANS) measurements on ternary blends was investigated. Results were analysed for two ternary systems of saturated hydrocarbon polymers by means of the multicomponent random phase approximation (RPA) as applied to the Flory-Huggins (FH) model. In one ternary blend-head-to-tail polypropylene, head-to-head polypropylene, and a saturated polyisoprene with 50% 3,4 content-the interaction coefficient X(T) had been determined independently by SANS measurements on binary blends for all three component pair combinations, permitting a direct test of internal consistency for the ternary RPA. The ternary blend data were found to agree well with RPA predictions despite some highly unusual behavior among its binaries, including negative values of X(T) for saturated hydrocarbon pairs. In the other ternary blend-three statistical copolymers of ethylene (E) and 1-butene (B) with different E-B compositions-values of X(T) were available for only two of the binaries, the third pair being immiscible over the range of available temperatures. The SANS data for the ternary blend were used to evaluate the interaction coefficient for the third pair. The interaction coefficients obtained were smaller than values calculated from previously assigned solubility parameters, a result that is disappointing although nonetheless still consistent with results obtained for other strongly interacting pairs.