The effects of 2 different prominently displayed appeals are examined in combination with 2 different prominently displayed confidentiality assurances on mail-back completion rates for census questionnaires. The experiment was carried out on a national probability sample of 30,000 household units, using occupant-addressed questionnaires that contained the content of the 1990 decennial census short form. Neither the benefits appeal nor the variation between 2 confidentiality assurances had a significant effect on completion rates. The mandatory appeal, however, significantly improved completion rates, even when it appeared on the envelope only.