Both environmental DNA (eDNA) and environmental RNA (eRNA) have been widely adopted for biodiversity assessment. While eDNA often persists longer in environments, eRNA offers a more current view of biological activities. In eRNA metabarcoding, extracted eRNA is reverse transcribed into complementary DNA (cDNA) for metabarcoding. However, the efficacy of various reverse transcription strategies has not been evaluated. Here we compared the biodiversity recovery efficiency of three strategies: random priming with hexamers, oligo(dT) priming, and taxa-specific priming using Mifish-U for fish in both high- and low-biodiversity regions. Our results demonstrate that reverse transcription strategies significantly impact biodiversity recovery. Random priming consistently detected the highest number of taxa in both low- and high-biodiversity regions. In low-biodiversity areas, oligo(dT) performed comparably to random hexamers; however, in high-biodiversity regions, random hexamers outperformed oligo(dT), particularly in recovering rare taxa. While taxa-specific priming was comparative to the other strategies for high-abundance taxa, it was less effective for rare taxa, thus limiting its utility for comprehensive biodiversity assessment. These differences are largely due to the multiple binding sites for random hexamers compared to the fewer or absent sites with oligo(dT) and taxa-specific primers under high eRNA degradation. Combining random hexamers and oligo(dT) significantly improved taxa recovery, especially for low-abundance species, supporting its best practice in eukaryotes. For prokaryotes or genes lacking polyadenylation, random priming is favored over taxa- or gene-specific priming. Collectively, these findings underscore the critical importance of selecting appropriate reverse transcription strategies in eRNA metabarcoding, with significant implications for effective biodiversity monitoring and conservation efforts.