Studies using controlled soil heating in the laboratory make it possible to understand and portray different scenarios of fire severity in the soil. However, some laboratory soil heating techniques do not maintain some field variables that are indispensable in heating and heat transmission through the soil and may underestimate the real severity of heating. This study aims to propose a new method for fire simulation, considering influence factors that are closer to the reality of natural fires. For that purpose, undisturbed samples were taken at 0–5 cm depth from Red-Yellow Ferralsol (n = 52) and Haplic Cambisol (n = 52) in the Vila Velha State Park – Paraná states - southern Brazil. The samples were experimentally heated in a conventional electric oven with internal air circulation based on three real field variables: oven temperature (100; 200; 300ºC), plant litter cover (absent and present), and soil moisture (<5 and 15–20 %), composing a 2x3x2 × 2 factorial scheme (n = 4). During heating, soil surface temperature was measured using thermocouples. The Kruskal-Wallis test was applied, followed by the Dunn test (p < 0.05), considering the asymmetric and non-normalized distribution of the variables. The final temperatures reached in both soils varied between 36.3 and 220.0°C, being reduced when moisture was higher, plant litter cover was present, and sand content was lower. In this test, therefore, the most influential factors in soil heating were the applied temperature and vegetation cover. The positive points of the proposed method are: 1) associating two or more factors in controlled experiments; 2) quantifying the magnitude of influence for individual or combined factors; 3) contributing to decision making regarding the proper moment to apply controlled fire, once the effects at different conditions of moisture, soil cover and type are identified; and 4) evaluating the realistic effects of low and medium intensity natural fire heating on soil properties.