Cost is a priority consideration in soil heavy metal investigation. For the correction of in-situ portable X-ray fluorescence (PXRF), traditionally-used linear regression (LR) is sensitive to outliers and cannot deal with the spatially varying influence of the relevant environmental factors. Moreover, each land-use type usually has a specific impact on the analysis quality of in-situ PXRF but was rarely incorporated into the correction model. This study first proposed robust geographically weighted regression with land-use types (RGWR-LUT) to correct in-situ PXRF Pb in an urban–rural transition area of Changzhou City, China. Next, its correction quality was compared with that of LR, LR with land-use types (LR-LUT), GWR, GWR with land-use types (GWR-LUT), and RGWR. Then, sequential Gaussian simulation with the corrected (by RGWR-LUT) in-situ PXRF as hard data (SGS-PXRF*) was proposed for spatial simulation of soil Pb. Finally, its spatial simulation quality was compared with that of SGS, SGS with the raw in-situ PXRF as hard data (SGS-PXRF), sequential Gaussian co-simulation with the raw in-situ PXRF as co-variable (CoSGS-PXRF), and CoSGS with the corrected (by RGWR-LUT) in-situ PXRF as co-variable (CoSGS-PXRF*). Results showed that: (i) spatially varying correction coefficient surface for in-situ PXRF Pb were obtained by RGWR-LUT; (ii) correction quality for in-situ PXRF Pb was in the order of RGWR-LUT (RI = 81.44%), GWR-LUT (RI = 72.1%), RGWR (RI = 67.83%), GWR (RI = 51.07%), LR-LUT (RI = 48.71%), and LR (RI = 22.72%); (iii) the corrected in-situ PXRF provided more information input than the raw in-situ PXRF for spatial simulation of soil Pb; (iv) SGS-PXRF* obtained the highest spatial simulation quality (RI = 62.5%). It is concluded that RGWR-LUT could effectively correct in-situ PXRF, and SGS-PXRF* could effectively incorporate the corrected in-situ PXRF for spatial simulation of soil heavy metals.