Web-based studies have become increasingly common in the social sciences, but have been rare in genetic epidemiology in general and twin studies in particular. We here review the methods, validity checks and preliminary correlational data from an on-line questionnaire collected from 2005–2008. During this time period, 44,112 individuals completed the questionnaire. This sample was 65.3% female, 85.4% 18 years or older, 72.0% Caucasian and had a mean educational level of 12.2 years. The sample included 609 twin, 333 sibling and 201 parent-offspring pairs as well as 342 dating partners, 313 ‘significant other’ pairs, 327 spouses and 2,316 friend pairs. A range of checks suggested low levels of invalid data. Correlations for personality, substance use and misuse, lifetime major depression, social attitudes, educational status, and height and weight were broadly similar to those obtained previously using conventional assessment methods. Web-based studies are a relatively easy and inexpensive way to ascertain large numbers of individuals, although obtaining twin pairs is more difficult, and female and monozygotic pairs are overrepresented. The sample is diverse and pair resemblance is generally similar to that obtained using interviews or mailed questionnaires.