Accurate identification of NAD-capped RNAs is essential for delineating their generation and biological function. Previous transcriptome-wide methods used to classify NAD-capped RNAs in eukaryotes contain inherent limitations that have hindered the accurate identification of NAD caps from eukaryotic RNAs. In this study, we introduce two orthogonal methods to identify NAD-capped RNAs more precisely. The first, NADcapPro, uses copper-free click chemistry and the second is an intramolecular ligation-based RNA circularization, circNC. Together, these methods resolve the limitations of previous methods and allowed us to discover unforeseen features of NAD-capped RNAs in budding yeast. Contrary to previous reports, we find that 1) cellular NAD-RNAs can be full-length and polyadenylated transcripts, 2) transcription start sites for NAD-capped and canonical m7G-capped RNAs can be different, and 3) NAD caps can be added subsequent to transcription initiation. Moreover, we uncovered a dichotomy of NAD-RNAs in translation where they are detected with mitochondrial ribosomes but minimally on cytoplasmic ribosomes indicating their propensity to be translated in mitochondria.