Even though increased accumulations of radiolabeled antisense DNAs compared to control DNAs are becoming a routine observation in cultured tumor cells, trustworthy evidence of tumor targeting in vivo by an antisense mechanism remains elusive. The goal of this study was to obtain convincing evidence of antisense tumor targeting in nude mice by using two different tumors and both intratumoral (i.t.) and intravenous (i.v.) administration of radiolabeled antisense and control sense DNAs. Both the MDR++ cell line KB-G2 and its parent MDR+ cell line KB-31 were used in this study. The antisense (AS) DNA was directed against the AUG start codon of the MDR1 mRNA and, along with the sense (S) control DNA, was a uniform phosphorothioate administered naked. In previous cell culture studies from our laboratories, the accumulation of this AS DNA was strikingly high in KB-G2 cells and only average in KB-31 cells, a fact we attribute to the 1000-fold higher expression by RT-PCR of MDR1 mRNA in the former cell line. In this study, both DNAs were radiolabeled with (99m)Tc via MAG3 and administered i.t. or i.v. at 1 microg (100 microCi) per animal 24 h prior to sacrifice and dissection in mice bearing thigh tumors of about 1 g. Following i.t. administration, no statistically significant differences (Student's t test, p < 0.05, N = 4) between the AS and S DNA biodistributions in normal tissues were observed except in the KB-G2 mice in which muscle levels were lower for the S control. In contrast, tumor levels in the KB-G2 animals were significantly higher for the AS DNA vs S DNA (14.7 vs 8.5% ID/g) while this difference (8.6 vs 4.3% ID/g) was insignificant in the KB-31 animals. The whole body images obtained just prior to sacrifice clearly show improved targeting of AS DNA vs S DNA in the KB-G2 but not the KB-31 animals. Calculations based on these results show that about 60 000 AS DNAs accumulated specifically (i.e. AS DNA - S DNA) per KB-G2 tumor cell following i.t. administration. When administered i.v. rather than i.t., higher tumor levels in KB-G2 animals compared to KB-31 were not observed, most likely because of the lower dosage reaching the tumors. When the KB-G2 and KB-31 results are combined, no statistically significant differences between the AS and S DNA biodistributions in normal tissues were observed except in blood in which S DNA levels were higher and in spleen in which they were lower. In contrast, tumor levels were significantly higher for the AS DNA vs S DNA (0.100 vs 0.063% ID/g). Calculations based on these results show that about 400 AS DNAs accumulated specifically per tumor cell following i.v. administration. Therefore evidence for tumor targeting in vivo by an antisense mechanism has been obtained in that statistically higher tumor accumulations of the (99m)Tc-AS DNA were observed compared to the control (99m)Tc-S DNA both following i.t. and i.v. administrations. The successful localization of AS DNA in tumor demonstrates that in vivo AS targeting of tumor is feasible although improvements in tumor delivery and normal tissue clearance are needed for practical antisense imaging.