This experiment was designed to determine the effects of under-dosing on the frequency of benzimidazole resistant allele in the nematode Teladorsagia circumcincta. Fenbendazole (FBZ) was tested at 1/32, 1/16, 1/8 and 1/4 of the recommended dose for sheep (5 mg/kg body weight). The fraction of the susceptible homozygote (SS), susceptible heterozygote (RS) and resistant homozygote (RR) genotypes were compared among FBZ dose groups to evaluate differences between SS and RS genotype selective advantage. Almost all SS genotype worms were eliminated by 1/4 of the FBZ recommended dose, whereas a significant fraction of the RS genotype worms survived treatment. The selective advantage was 4·5 times higher for the RS genotype. This selective advantage was determined at 1/4 of the manufacturer's recommended dose of FBZ. This value should be taken as an indictor of the selective advantage of RS over the SS genotype when lambs are under-dosed. A computer simulation was used to study the putative spread of anthelmintic resistance over a range of RS selective advantages (2, 4·5 and 10-fold), with two average sizes of individual host worm population (20 or 2000 worms/host) and two initial R allele frequencies (0·1% or 1%). In all situations, the lowest selective advantage of the RS genotype over the SS genotype was sufficient to promote the spread of resistance in susceptible populations. When the RS genotype had no selective advantage over the SS genotype, genetic drift almost always led to the loss of the R allele, except in the largest populations (average size = 2000 worms).