Ore grade estimation plays an important role in recoverable resource calculations and mining projects. Classic geostatistical estimation methods are based on two-point statistics and cannot make use of multiple-point and higher-order statistics of the data. Recent multiple-point geostatistical methods (MPS) use multiple-point and higher-order features of the spatial data to model more complex phenomena. MPS algorithms extract the higher-order features from a training image (TI), but in many applications obtaining a representative training image is difficult. Very few studies have been conducted on determination of the possible effects of these features on the results of ore grade estimation. In view of this, the present study introduced two new multiple-point interpolation methods based on the implicit Volterra series. The proposed methods directly extract multiple-point interactions of the available hard data and do not require training images. Therefore, they are suitable for applications like ore reserve estimation and spatial mappings, in which there is no proper training image, but more information and data is available than under-informed applications. To improve the generalization of implicit Volterra series, a regularized least-square and an epsilon-insensitive cost function were used to implicitly estimate the Volterra series coefficients in the reproducing kernel Hilbert space (RKHS). This study shows that in such cases, the implicit Volterra series is equivalent to least square support vector machines (LS-SVM) and support vector regression (SVR), respectively. Using these implicit Volterra series, two new multiple-point spatial interpolation methods were presented. Also, in this study multiple-point spatial templates were used to extract multiplicative features of the input data. Therefore, an algorithm was presented to search the 3D space by these templates. The applicability of the regularized implicit Volterra series in capturing the higher-order features of the spatial data is successfully evaluated in a synthetic example. An example iects the capability of the introduced implicit Volterra series in modeling the higher-order statistics of spatial data. In this example, using implicit Volterra series, higher-order components of a reference image, known as the crack image are computed. According to the results, it can be concluded that the introduced Volterra series can extract higher-order components of the spatial data, which is comparable to the orthogonal implicit Volterra series. Moreover, it can be concluded that the regularized Volterra series leads to a regular decay of the prediction error by increasing the order of the series. Also, using the regularized cost function, the overall accuracy of the predictions significantly improves. Also in another example, the generalization ability of the introduced methods is successfully examined in two real 2D and 3D mining examples at the Kuh-panj and Sarcheshmeh porphyry copper deposits. The results revealed that the introduced multiple-point methods could present better results than the ordinary and indicator kriging methods. Also according to the results, using an epsilon-insensitive cost function instead of a least square cost function improves the generalization ability of the implicit Volterra series