Behavioral variant of Frontotemporal Dementia (bvFTD) and Bipolar Disorder (BD) share overlapping symptoms, complicating diagnosis. BvFTD, especially linked to C9orf72 expansions, often mimics BD, highlighting the need for reliable biomarkers. This study aimed to differentiate bvFTD from BD using miRNA profiles in neural-enriched extracellular vesicles (NEVs). A cohort of 100 subjects was analyzed: 40 bvFTD (20 sporadic, 20 C9orf72 carriers), 40 BD, and 20 healthy controls. NEVs were isolated from plasma and profiled using real-time PCR. Among 754 miRNAs, 11 were significantly deregulated in bvFTD and BD. MiR-152-5p was downregulated in sporadic bvFTD, while let-7b, let-7e, miR-18b, and miR-142-5p were altered in C9orf72 carriers. BD patients showed distinct patterns in miR-331-5p, miR-335, and miR-345 compared to bvFTD. Bioinformatics analyses revealed that let-7e, let-7b, miR-18b, and miR-142-5p share common long non-coding RNA (lncRNA) targets, including XIST, NEAT1, and OIP5-AS1, suggesting their involvement in molecular networks relevant to C9orf72-related bvFTD. These miRNA signatures can differentiate bvFTD from BD, especially in C9orf72-related cases, and offer insights into disease pathways. Further research is needed to validate these findings and explore their clinical application.