Abstract This paper presents an experimental investigation of the sample thickness effect on the concurrent flame steady spread behavior. Two configurations, ceiling flame and floor flame, as the fundamental diffusion combustion problems and representative scenarios of horizontal flame spread, are concerned herein. PMMA samples with thickness ( δ ) of 2, 3, 4, 5, 7, 9 mm were employed including both thermally-thin and thermally-thick conditions. A wind tunnel was used to provide the uniform concurrent airflow ranging from 0.3 m/s to 1.2 m/s in 0.3 m/s intervals. Flame characteristics and basic parameters to quantify the flame steady spread process have been studied, including flame spread rate (FSR), equivalent pyrolysis length as well as heat flux feedback in both pyrolysis and preheating zones. Results show that FSR decreases with sample thickness, being more sensitive to sample thickness for ceiling flame configuration than that for floor flame configuration. Equivalent pyrolysis length increases with sample thickness and ceiling flame's equivalent pyrolysis length is larger than floor flame's. Convection feedback is dominant under concurrent airflow regardless of the flame configuration, airflow velocity or sample thickness. In pyrolysis zone, the total heat flux to fuel surface of ceiling flame configuration is larger than that of floor flame configuration for thin samples due to buoyancy, but it decreases rapidly as sample thickness increases. In preheating zone, the heat flux decreases in power law for ceiling flame configuration while a piece-wise fitting is found for floor flame configuration, which is interpreted based on their inherent flame characteristic difference hence the thermal boundary layer conditions between these two configurations. To analyze the sample thickness effect on concurrent flame spread, a characteristic solid temperature ( T * ) is proposed by theoretical analysis, as an essential parameter to be used in the concurrent flame spread model to represent the coupling effect of heat input in gas phase and in-depth heat loss into the solid phase (or namely thermal inertia controlled by a characteristic thermal penetration depth). The calculated non-dimensional characteristic solid temperature ( θ ) shows to be less sensitive to flame configuration and concurrent airflow velocity for relatively thicker samples because the thermal penetration depth will tend to approach a limit. A general trend of its variation with sample thickness ( δ ) can be employed to predict the heat loss into solid phase and its predictive capability is verified by the experimental results. This work advances the fundamental understanding of the sample thickness effect on concurrent flame spread mechanism.