The development of RNA sequencing (RNA-Seq) makes it possible for us to measure transcription at an unprecedented precision and throughput. However, challenges remain in understanding the source and distribution of the reads, modeling the transcript abundance and developing efficient computational methods. In this article, we develop a method to deal with the isoform expression estimation problem. The count of reads falling into a locus on the genome annotated with multiple isoforms is modeled as a Poisson variable. The expression of each individual isoform is estimated by solving a convex optimization problem and statistical inferences about the parameters are obtained from the posterior distribution by importance sampling. Our results show that isoform expression inference in RNA-Seq is possible by employing appropriate statistical methods.