Various sampling methods for Schistosoma mattheei cercariae, including sampling from different concentrations, were tested. The most satisfactory method consisted of mixing the cercarial suspension by pouring it to and fro between two cylinders and then sampling it by oral suction on a bulb pipette with a large aperture. The aliquots withdrawn were stored in bottles containing formol-saline for later microscopic examination. This method of sampling was tested extensively. The results showed that the variation between aliquots was larger than that expected for either a binomial or a Poisson distribution, indicating that the cercariae were not randomly distributed in the suspension. Moreover, significant variations occurred between the totals of series of aliquots, possibly indicating uncontrolled factors, e.g. in mixing. From these results, methods of making up doses of carcariae for infesting sheep and cattle were developed: The Sheep Method is used for doses of up to 15 000 cercariae and is therefore applicable to infestation of primates as well as sheep. Several series of aliquots are made up, some of which (one or more series per animal) are used for infesting the animals concerned (Infestation Series), while others are retained for estimating the infestation doses (Estimation and Additional Estimation Series). One Estimation and one Additional Estimation Series suffice for estimating the numbers of cercariae in up to eight Infestation Series drawn from the same cercarial suspension, which is mixed immediately before each series of aliquots is withdrawn. In the Bovine Method, which is used for doses of over 15000 cercariae, an Estimation Series and an Additional Estimation Series of aliquots are withdrawn from every dose of cercariae to be used for infesting an animal. Thus the number of cercariae is estimated separately for each animal. Formulae were compiled for these methods to determine the upper levels (with a specified probability) of the percentage errors in the estimated doses. These formulae were applied to the results and some expected percentage errors in the numbers of cercariae estimated by the two methods, calculated. A result for the probability distribution of aliquot counts under the assumption of randomness is proved for the first time.