RNA interference offers therapeutic opportunities for the clinical targeting of otherwise undruggable oncogenes. However RNAi can have off-target effects that considerably increase treatment risks. To manage these side effects and allow an easy subtraction of their activity in healthy tissues, we present here the TAG-RNAi approach where cells that are not designated targets do not have the mRNA tag. Using TAG-RNAi we first established the off-target signatures of three different siRNAs specific to the Cyclin D1 oncogene by RNA-sequencing of cultured cancer cells expressing a FLAG-HA-tagged-Cyclin D1. Then, by symmetrical allografts of tagged-cancer cells and untagged controls on the left and right flanks of model mice, we demonstrate that TAG-RNAi is a reliable approach to study the functional impact of any oncogene without off-target bias. Finally we show, as examples, that mutation-specific TAG-RNAi can be applied to downregulate two oncogenic mutants, KRAS-G12V or BRAF-V600E, while sparing the expression of the wild-type proteins. TAG-RNAi will thus avoid the traditional off-target limitations of RNAi in future experimental approaches.