The permeability of the margins of five dental cement seals was evaluated, by using 45Ca, after five different procedures for removing excess cement. The five dental cements varied significantly ( P < 0.001) in amount of penetration by 45Ca. From least penetration to greatest penetration, the order was (1) zinc oxide cement, (2) zinc phosphate cement No. 1, (3) silicophosphate cement No. 1, (4) silicophosphate cement No. 2, and (5) zinc phosphate cement No. 2. The five treatments also varied significantly ( P < 0.001). From least to greatest penetration, the order was (1) wiping the cement off the tooth and band prior to initial set, (2) hand scaling after complete set, (3) ultrasonic application directed only on the tooth surface, (4) ultrasonic application directed only on the band surface, and (5) ultrasonic application on tooth and band indiscriminately. The relationship among cements did not remain the same for all treatments; there was significant cement-treatment interaction ( P < 0.025). The buccal and lingual surfaces demonstrated less penetration than did the proximal surfaces ( P < 0.05). The two factors that emerged most clearly from this study were that the method of wiping or hand scaling the cement tended to prevent disruption of the integrity of the tooth-band cement bond, irrespective of the cement used, and, perhaps more important, that the zinc oxide-eugenol type of cement was able to withstand the effects of ultrasonic instrumentation.