The key step in the conversion of methane to polyolefins is the catalytic conversion of methanol to light olefins. The most recent formulations of a reaction mechanism for this process are based on the idea of a complex hydrocarbon-pool network, in which certain organic species in the zeolite pores are methylated and from which light olefins are eliminated. Two major mechanisms have been proposed to date—the paring mechanism and the side-chain mechanism—recently joined by a third, the alkene mechanism. Recently we succeeded in simulating a full catalytic cycle for the first of these in ZSM-5, with inclusion of the zeolite framework and contents. In this paper, we will investigate crucial reaction steps of the second proposal (the side-chain route) using both small and large zeolite cluster models of ZSM-5. The deprotonation step, which forms an exocyclic double bond, depends crucially on the number and positioning of the other methyl groups but also on steric effects that are typical for the zeolite lattice. Because of steric considerations, we find exocyclic bond formation in the ortho position to the geminal methyl group to be more favourable than exocyclic bond formation in the para position. The side-chain growth proceeds relatively easily but the major bottleneck is identified as subsequent de-alkylation to produce ethene. These results suggest that the current formulation of the side-chain route in ZSM-5 may actually be a deactivating route to coke precursors rather than an active ethene-producing hydrocarbon-pool route. Other routes may be operating in alternative zeotype materials like the silico-aluminophosphate SAPO-34.