In Brazil, water treatment sludge (WTS) production reaches 1.9 million m3/day (Calculated by the authors from the correlation between Brazilian drinking water production  and the percentage of WTS generated from raw water treatment [18–20].), mostly discharged in rivers, and minorly destined to landfills or sludge treatment plants. WTS reuse as a geotechnical material, a sustainable alternative to reduce environmental impacts, demands knowledge about WTS behavior. However, WTS characteristics vary for each WTP and along time for the same WTP, due to local geology, quality of raw water, type and quantity of coagulant, dewatering process, among others. Sampling procedures may also cause significant variation of tests results. To separate sampling errors from “real” variation of characteristics, representative samples for laboratorial tests must be composed using concepts of the Theory of Sampling (TOS). This paper evaluates the effects of sampling method and number of increments to compose the representative sample on the variability of water content and specific gravity of WTS generated at Cubatão WTP in São Paulo, Brazil. Results, valuated by analysis of variance methodology (ANOVA), indicated that both control parameters are influenced by the sampling method and also by the number of increments. In conclusion, the application of TOS concepts to compose representative samples is mandatory to obtain reliable WTS parameters.