While 9% of human pathogenic variants have an established effect on pre-mRNA splicing, it is suspected that an additional 20% of otherwise classified variants also affect splicing. Aberrant splicing includes disruption of splice sites or regulatory elements, or creation or strengthening of cryptic splice sites. For the majority of variants, it is poorly understood to what extent and how these may affect splicing. We have identified cryptic splicing in an unbiased manner. Three types of cryptic splicing were analyzed in the context of pathogenic variants in the acid α-glucosidase gene causing Pompe disease. These involved newly formed deep intronic or exonic cryptic splice sites, and a natural cryptic splice that was utilized due to weakening of a canonical splice site. Antisense oligonucleotides that targeted the identified cryptic splice sites repressed cryptic splicing at the expense of canonical splicing in all three cases, as shown by reverse-transcriptase-quantitative polymerase chain reaction analysis and by enhancement of acid α-glucosidase enzymatic activity. This argues for a competition model for available splice sites, including intact or weakened canonical sites and natural or newly formed cryptic sites. The pipeline described here can detect cryptic splicing and correct canonical splicing using antisense oligonucleotides to restore the gene defect.