Several tests for the detection of soy proteins in foods have been described in the literature, and some are commercially available. This article gives an overview of these methods and discusses the advantages and disadvantages of each individual method. Based on the conclusions of this inventory, an experimental approach was designed to improve the sensitivity of measuring soy protein in processed foods. The aimed sensitivity is 10 ppm (10 microg soy protein in 1 g solid sample), which is over 100-fold lower than presently available tests. The aimed sensitivity is this low because levels of food allergens at 10 ppm and above may provoke reactions in food allergic persons. Native soybean meal, soy protein isolate, soy protein concentrate, and textured soy flakes were used as test materials. Several extraction procedures were compared and a new method using high pH was selected. Polyclonal antibodies were raised in rabbits and goats, and immunopurified antibodies were used in sandwich and inhibition enzyme-linked immunosorbent assay (ELISA). Extraction at pH 12 resulted in good yields for all tested samples, both quantitatively (Bradford) and qualitatively by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Immunopurified rabbit antibodies against this extract used in a competition ELISA format resulted in a sensitive test with a detection limit of 0.02 microg/mL, corresponding to 0.4 microg/g (0.4 ppm) in food samples. Cross-reactivity with some main food ingredients was measured and appeared to be negative in all cases. The presently developed test is applicable for soy ingredients and soy-containing foods that are processed in different ways. The limit of quantitation is 1 ppm, which is an enormous improvement over earlier described methods.