The guidelines for good survey research advise to apply random sampling, as this makes it possible to generalise from the sample to the population. If the principles of probability sampling have been applied, it is always possible to compute unbiased estimates of population characteristics. This chapter compares surveys based on probability sampling with those based on non-probability sampling. The classical theory of probability sampling was more or less completed in 1952 when D. G. Horvitz and D. J. Thompson published their general theory for constructing unbiased (valid) estimates. The importance of probability sampling, as it makes it possible to compute valid (unbiased) estimates of population characteristics. Probability sampling also can suffer from problems affecting the representativity of its outcomes. One of the most important problems is nonresponse. Nonresponse occurs when people in the sample do not provide the requested information.