Fires in tunnels are a major concern due to the casualties they may cause. Therefore, forced ventilation is mandatory in long tunnels, despite the significant associated costs. In shorter tunnels, however, natural ventilation may be sufficient to comply with safety regulations. Accordingly, the analysis of natural fire smoke flow is relevant for tunnels shorter than 1000 m. This paper presents a computational investigation of the influence of the tunnel slope on the contamination of the cold lower layer with smoke and discusses how it impairs the user’s egress. Large-eddy simulations of the smoke propagation show three different regimes, namely, a quasi-horizontal tunnel behavior for a slope of 0.5%, a transitional behavior for slopes in the range of 1% to 5% and a quasi-forced ventilation behavior for a slope of 7%. The computational results are compared with the application of 1D equations to predict the upper layer temperature, the average mass flow rate, the upper layer mass flow rate, the upper layer velocity and the lower layer velocity. The distance from the fire to the location where the lower layer contamination with smoke starts is accurately predicted by the one-dimensional model for slopes of 2% and 3.5%. However, in the case of lower or higher slopes, the one-dimensional model performs poorly and needs further improvement.