Soil is a vital and dynamic component of our planet's ecosystem, and its health plays a fundamental role in sustaining life. Assessing soil health requires data and ways of interpreting the results. In this study, we focused on national and LUCAS datasets comparison for assessing and mapping soil health using pH as an indicator and a scoring function. Using mainland France as a pilot, we used the national RMQS dataset (2145 points) and the LUCAS dataset (2930 points) for the digital mapping of pH in combination with environmental covariates and a machine learning approach. A concept of soil districts (pedo-climatic zones) was implemented through stratification using soil, land use and climate types. We developed scoring functions based on mean, standard deviation, and Z scores using predicted pH values within each unique soil district. The “optimum is best” type of soil scoring function was used for health assessment. Our findings demonstrated that although spatial patterns of soil health levels were similar using both datasets, some regions still showed opposite results. These first set of results will be completed with other parameters such as SOC and developed within other countries. It also demonstrates the importance of appropriate choice of datasets in national soil health assessments.