Across a large variety of Mendelian disorders, ~50-75% of patients do not receive a genetic diagnosis by whole exome sequencing indicative of underlying disease-causing variants in non-coding regions. In contrast, whole genome sequencing facilitates the discovery of all genetic variants, but their sizeable number, coupled with a poor understanding of the non-coding genome, makes their prioritization challenging. Here, we demonstrate the power of transcriptome sequencing to provide a confirmed genetic diagnosis for 10% (5 of 48) of undiagnosed mitochondrial disease patients and identify strong candidate genes for patients remaining without diagnosis. We found a median of 1 aberrantly expressed gene, 5 aberrant splicing events, and 6 mono-allelically expressed rare variants in patient-derived fibroblasts and established disease-causing roles for each kind. Private exons often arose from sites that are weakly spliced in other individuals, providing an important clue for future variant prioritization. One such intronic exon-creating variant was found in three unrelated families in the complex I assembly factor TIMMDC1, which we consequently established as a novel disease-associated gene. In conclusion, our study expands the diagnostic tools for detecting non-exonic variants of Mendelian disorders and provides examples of intronic loss-of-function variants with pathological relevance.