To support future in situ/on-the-go applications using X-ray fluorescence (XRF) sensors for soil mapping, this study aimed at evaluating the XRF performance for predicting organic matter (OM), base saturation (V), and exchangeable (ex-) Mg, using a reduced analysis time (e.g., 4 s) in soil samples with different moisture contents. These attributes are considered secondary for XRF prediction because they do not present emission lines in the XRF spectrum. Ninety-nine soil samples acquired in two Brazilian agricultural fields were used. Soil samples with moisture content of 0, 5, 10, 15, 20, and 25 wt.% were measured by XRF under 4 s of dwell time. The results revealed that, despite the short dwell time, it was possible to obtain satisfactory predictions [residual prediction deviation (RPD) > 1.40] of V and ex-Mg in soil samples with up to 15 wt.% of moisture content. Satisfactory predictions of V were also possible with 20 and 25 wt.% of moisture content. Conversely, satisfactory predictions of OM were only possible in dried samples, yielding RPD of 1.60. Notwithstanding these promising results, for all studied attributes, the predictive performance gradually decreased as a function of water content in the soil. Nevertheless, we emphasize that the previously mentioned performance was obtained without the application of methods to mitigate the effect of moisture in spectral data [e.g., external parameter orthogonalization (EPO)]. Thus, methods to correct external effects on XRF data may lead to more accurate results, which highlights the necessity of further research to find the best method for mitigating the effect of soil moisture content in XRF spectra. This study emphasizes the potential of XRF for soil mapping in in situ analysis, being pioneering in showing that with a reduced scanning time it is possible to obtain satisfactory prediction performances for OM, V, and ex-Mg in wet samples.