Women recalled their sexual activity during pregnancy in telephone interviews occurring during and after the third trimester, as part of a study of preterm delivery in central North Carolina in 1996–97. This retrospective information was compared with prospective daily records for the same time periods in a subsample of 88 women. Relative to the daily records, women reported fewer episodes of intercourse and longer intervals since the most recent orgasm when they recalled this information in interviews. Recall of sexual activity over short, recent time periods during pregnancy appears to be a source of some misclassification. Reliance on retrospective reports may suffice for general measures of objective events, such as any vs. no intercourse over a short time period.