Many online panels suffer from methodological problems. One such problem is undercoverage. Only persons with Internet access can be included in the survey. No information can be collected of those without Internet access. Another problem can be self-selection or non-probability sampling. Under-coverage and non-probability sampling may lead to biased results. This chapter explains how it is possible to overcome under-coverage and self-selection errors. Three types of weighting procedures are described which can be applied to correct for lack of representativity. An alternative solution is to build an online panel on the basis of a probability sampling, using traditional recruitment procedures. Furthermore, under-coverage problems can be diminished by providing Internet access and equipment to all respondents where necessary. The LISS panel is described as a successful example of such a panel and is compared to other online studies and to a face-to-face study. In terms of representativity, it is close to the face-to-face survey and better than the other online surveys on five of the seven characteristics tested.