We develop a Flory mean-field theory for viral RNA (vRNA) molecules that extends the current RNA folding algorithms to include interactions between different sections of the secondary structure. The theory is applied to sequence-selective vRNA encapsidation. The dependence on sequence enters through a single parameter: the largest eigenvalue of the Kramers matrix of the branched polymer obtained by coarse graining the secondary structure. Differences between the work of encapsidation of vRNA molecules and of randomized isomers are found to be in the range of 20 k(B)T, more than sufficient to provide a strong bias in favor of vRNA encapsidation. The method is applied to a packaging competition experiment where large vRNA molecules compete for encapsidation with two smaller RNA species that together have the same nucleotide sequence as the large molecule. We encounter a substantial, generic free energy bias, that also is of the order of 20 k(B)T, in favor of encapsidating the single large RNA molecule. The bias is mainly the consequence of the fact that dividing up a large vRNA molecule involves the release of stored elastic energy. This provides an important, nonspecific mechanism for preferential encapsidation of single larger vRNA molecules over multiple smaller mRNA molecules with the same total number of nucleotides. The result is also consistent with recent RNA packaging competition experiments by Comas-Garcia et al.(1) Finally, the Flory method leads to the result that when two RNA molecules are copackaged, they are expected to remain segregated inside the capsid.