Since it was founded, INED has used a variety of sampling methods to carry out its surveys. In certain cases, all the statistical units eligible at the time of the survey are interviewed exhaustively. This is known as "a census". This method was used at INED for the 1998 survey in on resuscitating anaesthetists and is applied in each of the successive survey waves conducted since 1970 at Senegal's demographic surveillance sites. All the villagers in the survey area are interviewed.
However, due to the high logistical and financial costs of this type of operation, the majority of surveys do not cover the entire target population, but only a sample of it. Sample construction (or the "sampling plan") may be based on probability or nonprobability methods (also referred to as empirical methods). The choice between these methods depends on the financial and technical constraints and notably on the possibility of using a sampling frame. While nonprobability sampling methods (quota method in particular) were applied almost exclusively at INED until the 1970s, surveys based on a probability sampling plan have been increasingly used since then.

The sampling plan refers to the method of selecting the sample of survey respondents. Once the survey scope has been defined, two groups of methods can be used to constitute the sample.

Probability sampling designate a method by which each unit of the target population has a given probability, which is known or can be calculated in advance (before selection), of being included in the sample. Probability calculation techniques can then be used to make inferences for the entire population. INED has used this type of method in numerous surveys, for example the 1994 survey on "Les situations familiales et l'emploi" (family situations and employment), the 1996-1997 survey on "Devenir des enfants naturels" (fate of children born outside marriage) or again the "Biographies et entourage" (Event history and contact circle) survey in 2000-2001.

There are a variety of nonprobability sampling methods - quota method, itinerary method, case study method, etc. - used by INED in the past and up to the present day. Examples include the "Formation des couples" (union formation) survey of 1983-1984 or, more recently, the survey on "L'adoption dans 10 départements" (Adoption in 10 French départements) of 2003-2004.

For surveys on the general population, INED often uses INSEE's master sample. This is a list of dwellings drawn from the general population census, then updated using the directory of new housing. Probability sampling can then be implemented, as was the case for the ERFI survey in 2005.
The telephone directory has also been used to construct the sample for some surveys: for example, the "Proches et Parents" (local family circle) survey in 1990 and the ACSF (analysis of sexual behaviours in France) survey in 1992.
To address specific problems, INED has contributed to the development of original, highly complex sampling plans. For example, for surveys of the homeless conducted between 1995 and 1998, the sampling plan involved obtaining a sample of the homeless population via the services (meals, accommodation, etc.) that they used. INED began by identifying these services and so that their users could then be interviewed.
Similarly, in the "Famille et employeurs" (family and employers) survey of 2004-2005, the sampling plan coupled data collection from individuals with a survey of the establishment where they were employed.