The advent of ultra-high throughput sequencing technology has made it possible to directly sequence large numbers of mRNA fragments, obtaining a direct measure of mRNA abundance.  Restriction enzyme fragmentation reliably produces a single sequenced fragment per mRNA, providing a direct measure of digital gene expression from tag counts.  However, genes in complex organisms may encode multiple mRNAs, called isoforms which result in different proteins.   Since each restriction site can belong to multiple isoforms, the counts must be partitioned among the isoforms to derive accurate isoform counts.  We combine a model for fragmentation with methodology for contingency table analysis to estimate isoform abundance taking into consideration differing probabilities of tag formation failure for each isoform.