Retrospective assessment of tobacco use underlies much of the data collected in epidemiological and genetic epidemiological research. Although individuals are asked to report lifetime tobacco use for periods spanning months to decades, the test-retest reliability intervals of the instruments often span only a few weeks to several months. The present analyses examined the test-retest reliability of retrospective tobacco use measures, including details of first use, circumstances of first use, and initial subjective reactions. The questions were part of the Lifetime Tobacco Use Questionnaire (LTUQ), a Web-based questionnaire designed to assess use of most forms of tobacco or nicotine retrospectively across the lifespan. A convenience sample of 236 men and women with history of tobacco use (Time 1 mean age, 44.9 years; 74.2% females; 75.1% regular monthly tobacco use) responded verifiably to invitations to self-administer the LTUQ two times, 2 years apart. Test-retest reliability analyses reflected high reliability for salient tobacco-use questions. Acceptable levels of reliability were observed for initial subjective reactions to smoking, if the scaled response options were dichotomized. Few differences in the reliability of recall were apparent between sexes and between age groups. These results indicate that recall of important tobacco use information can form a reliable basis for research.