Retrotransposons, also called Class I transposable elements, and their fragments can constitute upwards of 50% of higher plant genomes. The most numerous subclass, the Long Terminal Repeat (LTR) retrotransposons, resemble retroviruses in that they are transcribed from promoters in the LTRs and encode proteins for packaging of the transcripts, their reverse transcription into cDNA, and integration of the cDNA back into the genome. In addition, the genome contains non-autonomous LTR retrotransposons, which parasitize the proteins of autonomous elements for their replication and propagation. A major question for both retrotransposons and retroviruses is how the two conflicting roles for their transcripts, translation and reverse transcription, are managed. Here, we show that the copia-like BARE retrotransposon, despite its simple organization into only one open reading frame, produces three distinct classes of transcripts, one that is capped, polyadenylated, and translated but which cannot be reverse transcribed and the other, which is not capped or polyadenylated, but is destined for packaging and ultimate reverse transcription. The third class is capped, polyadenylated, and spliced to favor produce a subgenomic RNA encoding only Gag, the capsid protein that forms the virus like particles (VLPs). Moreover, the BARE2 subfamily, which cannot synthesize Gag and is parasitic on BARE1, does not produce the spliced sub-genomic RNA for translation but does make the replication competent transcripts, which are packaged into BARE1 particles. To our knowledge, this is first demonstration of distinct RNA pools for translation and transcription for any retrotransposon.