We present a field-based approach to detect and monitor insects with resistance to insecticidal toxins produced by transgenic plants. Our objective is to estimate the phenotypic frequency of resistance in a population by relating the densities of insects on genetically transformed plants to densities on nontransformed plants. We focus on European corn borer, Ostrinia nubilalis (Hübner), in sweet corn, Zea mays L., expressing Cry1Ab from Bacillus thuringiensis subsp. kurstaki Berliner to illustrate principles underlying the method. The probability of detecting one or more rare, resistant larvae depends on sample size, the density of larvae on nontransformed plants, and an assumed frequency of resistant phenotypes in a given population. Probability of detection increases with increases in sample size, background density, or the frequency of resistant individuals. Following binomial probability theory, if a frequency of 10(-4) is expected, 10(3)-10(4) samples must be collected from a B. thuringiensis (Bt) crop to have at least a 95% probability of locating one or more resistant larvae. In-field screens using transgenic crops have several advantages over traditional laboratory-based methods, including exposure to a large number of feral insects, discrimination of resistant individuals based on Bt dosages expressed in the field, incorporation of natural and Bt-induced mortality factors, simultaneous monitoring for more than one insect species, and ease of use. The approach is amenable to field survey crews working in research, extension, and within the seed corn industry. Estimates of the phenotypic frequency of resistance from the in-field screen can be useful for estimating initial frequency of resistant alleles. Bayesian statistical methods are outlined to estimate phenotype frequencies, allele frequencies, and associated confidence intervals from field data. Results of the approach are discussed relative to existing complementary methods currently available for O. nubilalis and corn earworm, Helicoverpa zea (Boddie).