Unintended pregnancies can have serious health, social, and economic consequences. Such pregnancies may be unwanted (a baby is not wanted at any time) or mistimed, yet wanted (a baby is wanted eventually). Intended pregnancies are those conceived when desired. Reproductive health survey respondents' understanding of these concepts and validity of survey results may be affected by question order and wording. Using a randomized crossover design, National Survey of Family Growth (NSFG) and Demographic and Health Survey (DHS) intendedness questions were asked in a 1993 survey of Arizona women aged 18–44 years. Of 2, 352 ever-pregnant respondents, 25% gave discordant responses to DHS and NSFG questions about the most recent pregnancy. Age, marital status, household income, education, parity, time since pregnancy, and outcome of pregnancy were significantly predictive of discordant responses. DHS and NSFG questions yielded similar prevalence estimates of intendedness and wantedness; but young, unmarried respondents gave more “mistimed” responses on whichever question was asked later. Classifying pregnancies as intended, mistimed, or unwanted may be a problem for women who have not decided on lifetime reproductive preferences. Approaches to improving survey validity include addressing ambivalence, clarifying the definition of “unwanted”, and for young, unmarried women, not attempting to classify unintended pregnancies as mistimed or unwanted. Am J Epidemiol 1997; 145: 810–16.