Global demand for soil information has led to investigations that have adopted ways to estimate soil attributes quickly and effectively. In this context, magnetic susceptibility (χ) has gained prominence because it is a technique capable of estimating other attributes that are more difficult to acquire. This study aimed to (a) evaluate the performance of χ for the prediction of sand, silt, clay, hue, hematite/(hematite + goethite) ratio, Fe content of pedogenic iron oxides, and remaining phosphorus and (b) develop maps of χ, soil attributes and attributes predicted by χ in the state of Rio Grande do Sul (RS), Brazil. Here, 198 soil samples under forest and native pasture were used for testing the potential of χ as a predictive technique, separating the data into calibration (nc = 149) and validation sets (nv = 49). Linear regression was used to obtain the pedotransfer equations according to soil classes and lithology. To visualize the distribution of the values of χ and other soil attributes in RS, maps were made with the real values of χ and the real and estimated values of soil attributes. The great range of the χ values and related attributes was associated with the lithological and pedological influence, allowing the construction of predictive models that encompass a large gradient of χ. In the predictions made in groups, the attributes of Oxisols and Ultisols were best estimated by χ; however, among the lithology groups, the extrusive igneous rocks stood out.