Formalin-fixed paraffin-embedded (FFPE) tissues are commonly used both clinically and in forensic pathology. Recently, noncoding RNA (ncRNA) has attracted interest among molecular medical researchers. However, it remains unclear whether newly identified ncRNAs, such as long noncoding RNA (lncRNA) and circular RNA (circRNA), remain stable for downstream molecular analysis in FFPE tissues. Here, we assessed the feasibility of using autoptic FFPE brain tissues from eight individuals to perform quantitative molecular analyses. Selected RNA targets (9 mRNAs and 15 ncRNAs) with different amplicon lengths were studied by RT-qPCR in paired fresh and FFPE specimens. For RNA quality assessment, RNA purity and yield were comparable between the two sample cohorts; however, the RNA integrity number decreased significantly during FFPE sampling. Amplification efficiency also displayed certain variability related with amplicon length and RNA species. We found molecular evidence that short amplicons of mRNA, lncRNA, and circRNA were amplified more efficiently than long amplicons. With the assistance of RefFinder, 5S, SNORD48, miR-103a, and miR-125b were selected as reference genes given their high stability. After normalization, we found that short amplicon markers (e.g., ACTB mRNA and MALAT1 lncRNA) exhibited high consistency of quantification in paired fresh/FFPE samples. In particular, circRNAs (XPO1, HIPK3, and TMEM56) presented relatively consistent and stable expression profiles in FFPE tissues compared with their corresponding linear transcripts. Additionally, we evaluated the influence of prolonged storage time on the amplification of gene transcripts and found that short amplicons still work effectively in archived FFPE biospecimens. In conclusion, our findings demonstrate the possibility of performing accurate quantitative analysis of ncRNAs using short amplicons and standardized RT-qPCR assays in autopsy-derived FFPE samples.