Biomass to be composted is often very heterogeneous and collection of representative samples for determination of compost properties is therefore difficult, especially under full-scale conditions. During full-scale composting different biomasses in the amount of 10–100 tons are mixed, yielding a very heterogeneous mixture. Final sample size for compost property determination is usually a few grams compared with compost pile masses of hundreds of tons. Desired sample particle size is about 1 mm, while compost particle size ranges from 5 to 50 cm. This study focuses on the development of a strategy for sampling under full-scale conditions for minimum measurement uncertainty based on selected material properties. Optimization was conducted considering multiple parameters, such as number of pile turnings before sampling, number of samples collected, sample mass, sample homogenization, particle size reduction and number of replicate measurements. Measurement uncertainty was evaluated using water content, inorganic matter content and nutrient (nitrogen, phosphorus) content. For each parameter measurement variability was determined as a function of sampling strategy and used to identify optimal sampling strategy.