Ice sheet inception at the onset of the last glacial epoch, ca. 115 kyr BP, is not well understood. To cause the Northern Hemisphere to descend into Ice Age conditions, internal climate system feedbacks must have acted to amplify relatively low-amplitude orbital cooling trends. While a number of plausible internal mechanisms have been proposed for this internal amplification, the dramatic continental ice sheet expansion that is recorded in the marine sediment record is an enduring mystery. Global sea level lowering of at least 20 m and as much as 75 m over 10 kyr appears typical of the early stages of glacial inception. This rate of ice sheet growth has not been freely captured in either glaciologic or climatic models. Is it because we do not adequately understand the relevant internal feedbacks, or fail to capture them in models? Is it simply related to technical model limitations, e.g. topographic resolution? This paper explores the premise that model resolution poses a critical limitation on simulations of ice sheet nucleation. I present a series of ice sheet nucleation simulations in North America and Eurasia based on an ice sheet model that includes parameterizations of km-scale topographic detail, snow/ice accumulation and melt, and glacier dynamics. Model climate is based on perturbations to modern observational climatology. Even with this topographic detail and with simulations long enough to allow ice to advect downslope and spread over the terrain, it remains difficult to induce a large-scale glacierization in both North America and Eurasia; severe climate perturbations are required. Cold temperatures need to be accompanied by substantial increases in precipitation in the North Atlantic and polar regions in order to accrue continental ice at the rate demanded by sea level reconstructions. The simulations presented here are too simple to separate climate perturbations from internal system feedbacks, but with modern precipitation rates, a combined climate cooling/net positive feedback of more than 8°C is required to establish large-scale ice sheets in <10 kyr. With significant regional increases in moisture (50–100% more precipitation), a net mid-latitude cooling of 5–6°C is required to achieve a sea level lowering of more than 20 m in 10 kyr. I suggest that some reorganization in North Atlantic and Arctic climate in the late Eemian may be necessary to facilitate these substantial climate shifts.