Congenital myopathies (CM) are a group of clinically heterogenous disorders, defined by muscle histopathology, and with overlapping genetic etiologies. In the Beggs Laboratory at Boston Children’s Hospital, approximately 30% of patients with CM remain without a genetic diagnosis. Here, we evaluate the utility of RNA sequencing (RNAseq) to diagnose additional cases through identification of pathogenic variants that impact splicing and/or lead to mono-allelic expression of heterozygous variants. We performed RNAseq on 145 skeletal muscle biopsies, of which 103 lacked genetic diagnosis. Thirty-eight were previously diagnosed, and four were unaffected tissues from healthy donors. We utilized the Detection of RNA Outliers Pipeline (DROP) bioinformatic tool, which includes FRASER2, OUTRIDER, and Mono Allelic Expression (MAE) modules. Outcomes from these modules were filtered using a comprehensive list of skeletal muscle genes and cases were individually reviewed accounting their clinicopathological phenotype and inheritance model. One of the main difficulties of analyzing DROP outputs is the large amount of data produced per patient, even after statistical filtering. To address this hurdle, we characterized each DROP output at the cohort level to identify global patterns of splicing and expression abnormalities. Furthermore, we used XGBoost, a machine learning algorithm, to identify myopathy subtype-specific transcriptomic signatures in clinical diagnoses where we had at least 5 samples. Results, mostly in combination with exome sequencing data, led to 18 (17.5%) solved cases and 5 cases (4.8%) with strong candidate genes identified. Of solved cases, 16 (89%) were found with FRASER2, while 2 (11%) were found with MAE. These results demonstrate the utility of RNAseq to identify variants that lead to disease-causing aberrant splicing and advance our goal to increase diagnostic yield and guide gene discovery in congenital myopathies.