Recent technological advances have enabled large RNA-Seq experiments to be performed on multiple related samples. For example, the Geuvadis project has examined expression in hundreds of individuals for the purpose of identifying eQTLs that are single nucleotide polymorphisms associated with expression levels of genes. Underlying such experiments is a tradeoff between the number of samples that are sequenced, and the depth at which each sample can be sequenced. Pooling of samples can improve the overall accuracy of abundance estimates, yet such averaging defeats the purpose of the experiment: namely the analysis of differences between samples. We describe a computational approach to pooling of multiple RNA-Seq experiments that allows for borrowing of information between samples.