Upward flame spread has the same propagating direction with air flow and buoyancy, and features as the most hazardous fire case in all flame spread configurations. It has been a long time for fire researchers to find a simple and effective method to evaluate upward flame spread behaviors, especially for different materials and sample sizes. The aim of this work was motivated by the research of sample width and thickness effects on upward flame spread behavior, including flame spread rate during acceleration propagation for different sample thickness and width, theoretical global mass loss prediction based on Emmons's hypothesis, and dimensionless flame height scaling with dimensionless heat release rate for steady stage burning. Four kinds of sample thicknesses were selected, including 1.7, 3.5, 5, and 7mm. For each kind of thickness, six sample widths ranging from 40 to 90mm were prepared. To eliminate the side flame spreading effects, one set of contrast experiments with sample sides sealed was also performed, by which way flame could only spread along sample front surface and flame propagation was inhibited along both sides. Based on Emmons's hypothesis, a method for calculation of global mass loss rate was developed. Theoretical global mass loss rate over pyrolysis surface of upward flame spread configurations was calculated and could fit the experimental data well. Finally, a dimensionless heat-release rate for wall flames of different sample sizes was used to scale the dimensionless flame height with a power-law exponent 0.58. The results of this study have implications concerning designs for high-rise building fire safety problems and can help to get better understandings of upward flame spread mechanism from aspects of heat and mass transfer.