Fire causes changes in soil moisture content (MC) and also in other soil properties depending on the maximum temperature reached. However, after fire, MC is partially rapidly restored due to re-equilibrium with air moisture or following rainfall, which in turn affects the near infrared (NIR) spectra of soil. The degree to which MC is restored depends on other soil properties, which are also affected by heating. We studied the effect of soil moisture on the estimation of the maximum temperature reached (MTR) in the soil using NIR spectroscopy. Different pre-treatments of burned samples were applied to modify MC. Different models relating NIR spectra with MTR were developed using partial least squares (PLS) regressions for the different pre-treatments. Models were successfully leave-one-out cross-validated with r 2 greater than 0.982 and with the root mean square error of cross-validation (RMSECV) smaller than 28.2 °C. These models were used to predict the MTR for an independent set of soil samples heated at seven temperatures of 100, 200, 300, 400, 500, 600 and 700 °C. The MC of this set of samples was also modified using similar pre-treatments. Accurate estimations of MTR were obtained when the pre-treatment of samples used was similar as that of the set of samples used in the calibration model. Although MC of burned soils can affect MTR estimations, MTR can be estimated using NIR with high accuracy. Each heating temperature produced a gradient of changes leading to different NIR spectra that was not solely due to changes in soil moisture. This fact allows MTR to be estimated from NIR spectra.