A new experimental technique has been developed for determining the compositional distribution of ethylene copolymers and blends of ethylene homopolymers and copolymers. The method is similar to the turbidimetric titration procedure, but with crystalline precipitation of the various structural species effected by lowering of the temperature. Experimentally, a Brice-Phoenix light-scattering photometer, equipped with a stirred heated cylindrical cell is used as a turbidimeter for measuring the light transmission through a liquid system. The decreasing light transmission of a 0.01% solution in a fixed ratio of solvent and nonsolvent of α-chloronaphthalene and dimethyl phthalate, with a temperature drop of 2°C./min., is plotted against temperature on an X-Y recorder. The more highly branched molecules, that is, those containing higher comonomer content, are the less crystalline, hence are more soluble and precipitate at the lower temperatures. Compositional distribution is estimated from the cooling curves by two procedures: (1) the broader temperature range over which a nonuniform resin precipitates out of solution is compared to that of a more uniform resin, and (2) the higher initial cloud point of a mixture or nonuniform copolymer is compared to the cloud point–comonomer content relationship developed for uniform resins. The effects of molecular weight and molecular weight distribution are shown to be small relative to the effects of structural distribution. The compositional distribution of the ethylenepropylene copolymers studied varies from narrow to broad, depending on the specific coordination catalyst and polymerization method used. Ethylene–vinyl acetate copolymers, on the other hand, are uniform because of the 1:1 reactivity ratios of ethylene and vinyl acetate. Synthetic mixtures of ethylene–vinyl acetate resins of varying vinyl acetate content exhibit the expected nonuniformity. This technique has also been successfully applied to mixtures of high- and low-density polyethylenes, further demonstrating the utility of this new rapid tool for structural characterization. Experimental time is about 2 hr.