We present the results of a study by Dessart et al. (2012), where we performed stellar collapse simulations of proposed long-duration γ-ray burst (LGRB) progenitor models and assessed the prospects for black hole formation. We find that many of the proposed LGRB candidates in Woosley & Heger (2006) have core structures similar to garden-variety core-collapse supernova progenitors and thus are not expected to form black holes, which is a key ingredient of the collapsar model of LGRBs. The small fraction of proposed progenitors that are compact enough to form black holes have fast rotating iron cores, making them prone to a magneto-rotational explosion and the formation of a proto-magnetar rather than a black hole. This leads us to our take-home message, that one must consider the iron-core structure (eg. ρ(r), Ω(r)) of evolved massive stars before making assumptions on the central engine of LGRBs.