The occurrence of anthelmintic resistance to levamisole, albendazole, ivermectin and moxidectin was investigated in cattle from 10 farms located in São Paulo State, Brazil, using two techniques for counting eggs in faeces: McMaster with a sensitivity of 50 eggs per gram (EPG) and FLOTAC with a sensitivity of two EPG. We also evaluated the use of different mathematical and test design approaches to determine the efficacy of the anthelmintic treatments: one formula/design that compares post-treatment arithmetic mean EPG counts for the treated and control groups (FECRT1) and two methods to analyse data from pre- and post-treatment EPG counts in the same group (FECRT2 and FECRT3, respectively). Treatment groups received either ivermectin (0.2 mg/kg of body weight (BW); moxidectin (0.2 mg/kg BW); albendazole (2.5 mg/kg BW); levamisole (4.7 mg/kg BW); or no treatment (control group). The number of animals in each group ranged from 8 to 11. Faecal samples from each animal were collected 2 days before the treatment and again 10 and 28 days post-treatment. The FEC reduction (FECR) confidence intervals were usually wider when based on data obtained using the McMaster method than when data were obtained using the FLOTAC method. Efficacy estimated from pre- and post-treatment EPG counts in the same group presented smaller confidence intervals. Ivermectin proved to be totally ineffective in all herds evaluated. Cooperia spp. was the major parasite displaying resistance, followed by Haemonchus spp. The results also indicated the presence of Oesophagostomum spp. and Trichostrongylus spp., meaning they, too, were resistant to ivermectin. Resistance to moxidectin was found on nine of the 10 farms investigated; however, only three farms had previously used moxidectin. In contrast, albendazole and levamisole demonstrated high efficacy on the majority of farms. In surveys for anthelmintic resistance in cattle, the use of a diagnostic method with higher sensitivity to detect eggs is recommended, as is the case with the FLOTAC method. This study indicates that by using techniques with high sensitivity and by testing the same animals pre- and post-treatment, good precision can be achieved with group sizes from 8 to 11 animals.