The use of antibiotics and chemotherapeutics in animal husbandry has led to the occurrence of veterinary drug residues in all types of food of animal origin. Due to the specification of toxicologically based maximum residue levels for a large number of substances, existing control strategies need even faster and more sensitive methods to meet new and more rigorous regulations. The applicability of an immunosensor device for biospecific interaction analysis was investigated and the development of an assay for analysis of sulfamethazine (SMZ) in milk is described. SMZ was covalently immobilized to a carboxymethyldextran-modified gold film. Spiked samples with known concentrations of SMZ were prepared in HBS buffer and skim and raw milk for construction of standard curves. Polyclonal antibodies against SMZ were added to the sample and the immobilized surface was used to determine the amount of free antibodies by surface plasmon resonance detection. After each measurement the surface was regenerated by (\mathrm{NaOH}) and (\mathrm{HCl}). In milk, the mean relative standard deviation of the assay was approximately (2 %) and the limit of detection less than (1 \mathrm{ppb}). By introduction of a secondary sheep anti-rabbit antibody, the use of specific antibody could be reduced. Milk samples from the individual cow, herd, and tanker levels were analyzed and the relative standard deviations within each sample category were (4.4,2.4), and (2.2 %), respectively. The effect of some potential interferences, e.g., high somatic cells, bacterial contamination, and preservatives, was investigated. The results were not influenced in such a way that the risk for so-called false-positive findings was obvious. 1995 Academic Press, Inc.