Antibodies to polyethylene glycol (PEG) were raised in rabbits by immunization with monomethoxy polyethylene glycol modified ovalbumin (OA), bovine superoxide dismutase (SOD), and ragweed pollen extract (Rag), given in Freund's complete adjuvant (FCA). Immunogenicity depended on the nature of the protein and the degree of modification. With modified OA, in the presence of FCA, the majority of animals showed an anti-PEG response. With modified SOD and Rag only a small proportion of animals responded. In the absence of FCA, modified OA, given s.c., did not elicit any anti-PEG antibody response in rabbits and only a weak response in mice. PEG of MW 10,000 and 100,000 given in FCA was found nonimmunogenic in rabbits, and PEG of MW 5.9 X 10(6), given s.c. to mice, showed no or very poor immunogenic properties. Gel diffusion, heterologous passive anaphylaxis and passive hemagglutination were used to demonstrate anti-PEG antibodies raised to PEG-modified proteins. Specificity was confirmed by hapten inhibition of precipitation, inhibition of passive hemagglutination and cross-reactivity tests. PEG of MW greater than or equal to 4,000 produced specific precipitates, smaller molecules acted as monovalent haptens. From hapten inhibition of precipitation by PEG of MW 300 it appears that the antigenic determinant of PEG may be a sequence of 6-7 -CH2CH2O-units. Anti-PEG antibodies can be used analytically. By gel diffusion, Peg was detected in minimal concentrations of 0.1-1 microgram/ml. The clinical relevance of these findings with regard to therapy with PEG-modified enzymes and allergens in humans remains to be established.