This study aimed to evaluate the limit of detection of Toxocara canis larvae in experimentally contaminated commercial bovine milk samples, based on a centrifuge-sedimentation technique. Firstly, bovine milk (whole and skim) samples were contaminated with 50 T. canis larvae in order to evaluate the interference of milk fat with the recovery of the larvae. Next, the effects of 10% formalin (100 ?L), ether (100 ?L), and a combination of both solutions on the recovery of the larvae was examined. Thereafter, the limit of detection of the larvae was determined using the solution (from step 2) considered optimal for degreasing the milk sample. Samples were contaminated with aliquots of 1, 5, 10, 25, and 50 larvae. For each milk sample (1.0 mL), 15 repetitions were analysed. The recovery of the larvae from the skim milk samples was higher (p = 0.0031) than that from the whole milk samples. No significant difference (p = 0.5681) was observed with regard to the percentage of recovered larvae when comparing the degreasing solutions. Nevertheless, the formalin-ether combination was more efficient for recovering the larvae (73.1%) than ether (71.9%), formalin (67.6%), and pure whole milk (70.0%). Concerning the limit of detection (using formalin-ether), all the samples contaminated with 5, 10, 25, and 50 larvae tested positive (minimum: 62.7%). Of the samples contaminated with a single larva, 66.7% tested positive. These results suggest that the centrifugation-sedimentation technique may be useful for recovering larvae of Toxocara spp. in naturally or experimentally contaminated milk samples obtained from a wide range of animal species.