The superficial description in biomedical journals of sampling methods used in epidemiological studies of the prevalence of some diseases can be attributed to shallow knowledge of basic sampling techniques. The population of interest in most community surveys is usually very large and resources and time available limited, so that researchers have little or no choice but to study a sample of the population. One of the basic principles of sampling is the avoidance of bias, guaranteed by taking a random sample. But the term 'random sample' has often been misinterpreted as synonymous with 'haphazard sample', taking a sample without a definite pattern. It is re-emphasised that a random sample is a probability sample that gives every unit in the population a known probability of being selected in the sample. The procedures for taking a random sample for a nationwide study in the Kingdom of Saudi Arabia are not easy because of the structure of the population, and therefore require more complex sampling methods like the stratified cluster sampling. It is also necessary in a stratified sample to calculate estimated persons affected by a condition for each selected subgroup of the population before obtaining the overall prevalence rate. A proper understanding and use of appropriate sampling techniques is most likely to result in the most desired representative sample, and guarantees that some underlying assumptions for inferential statistics will be satisfied.