We examine the role of the condensing agent in the formation of polyelectrolyte bundles, via grandcanonical Monte Carlo simulations. Following recent experiments we use linear, rigid divalent cations of various lengths to induce condensation. Our results clarify and explain the experimental results for short cations. For longer cations we observe novel condensation behavior owing to alignment of the cations. We also study the role of the polyelectrolyte surface charge density, and find a nonmonotonic variation in bundle stability. This nonmonotonicity captures two trends that have been observed in separate experiments. [1,4] or the fd and M13 virus [2], they aggregate into dense, hexagonally coordinated bundles. The counterintuitive nature of the effective attraction implied by bundle formation, in combination with the biological importance of the phenomenon, have made it the topic of numerous studies over the past decades (see Refs. [5,6,7,8,9, 10] and references therein). There is now general agreement that strong electrostatic correlations are a crucial condition [9,11], in accordance with the experimental observation that most condensing agents carry a multivalent charge. However, other factors that affect the tendency of a system to exhibit condensation are much less well established, mainly because these factors are difficult to disentangle in both experiments and theory.Even for a simple ionic condensing agent, its interaction strength with the polyelectrolyte is affected not only by its valency, but also by its size [12] and by the surface charge density of the polyelectrolyte. In addition, the ionic concentration plays an important role, as it determines the stabilizing osmotic pressure exerted on the bundle by the surrounding solution [2, 4] and also controls the strength of entropic effects such as counterion release and depletion interactions [5]. Thus, even a systematic variation of the ion size can yield results that are difficult to interpret, as it alters both the binding of the ion to the polyelectrolyte and the osmotic pressure of the solution. Likewise, variation of the polyelectrolyte charge not only affects ionic binding, but also the direct electrostatic repulsion between polyelectrolytes. Accordingly, only an integrated approach can resolve the true origin of observed trends in the aggregation behavior.There is a remarkable dearth of such studies. Experimentally, there are technical limitations. For example, examination of the effect of counterion size is hampered by the uncertainty in measuring hydrated ion sizes [13] and variation of counterion valency often entails the simultaneous variation of other ionic properties. Furthermore, it is difficult to measure the ionic concentration within the aggregate and hence to assess the osmotic effects arising from a concentration imbalance with the bulk solution. This osmotic stress is also often ignored in computational and theoretical studies. Inspired by two recent experimental studies [13,14], in this Letter we aim to obtain a more complete u...