ASER samples households, not children. All children aged 3-16 who reside in sampled households are included in the survey. Enrolment information is recorded for all these children, while basic learning levels are assessed for children aged 5-16.

In each district, 30 villages are sampled from the most recent Census village directory using Probability Proportional to Size (PPS), a sampling technique that is commonly used for large scale surveys. 20 households are randomly sampled in each village. This generates a total of 600 sampled households in each rural district or about 3,00,000 households at the national level.

The sample design employs a rotating panel of villages. Each year, 10 villages from three years ago are dropped and 10 new villages are added. For example, in ASER 2010, 10 villages from ASER 2007 were dropped, 10 villages from 2008 and 2009 were retained and 10 new villages from the census village directory of 2001 were added. This strategy generates a representative picture of each district.