In the early 70s, the soil biologist Marcel Bouché classified French earthworms species into defined ecological categories. This classification system was immensely successful and is still widely used to describe earthworm functional groups even outside of Europe. Bouché used morpho-anatomical traits to differentiate three main categories: epigeic, anecic and endogeic. However, the way species are assigned to a category was not explicitly described in Bouché’s work. Thus, some earthworm species can still be assigned to two categories depending on the way researchers interpret Bouché’s description. To solve these issues and avoid unnecessary controversies, we applied PCA and random forest models to the seminal data of Marcel Bouché (earthworm morpho-anatomical traits). Their assignment to Bouché’s three main categories allowed us to statistically redefine the different categories and determine which traits are the most influential. We found that the three main traits were skin pigmentation (from none to black), body length (mean of the minimal and maximal values) and skin coloration (yes or no), followed by 10 other morphological and anatomical traits. We then used this approach to assign a likely category to all of the species studied by Bouché, resulting in a new triangular graph including other categories such as epi-anecic, endo-anecic, epi-endogeic and intermediate. Finally, we calculated the percentage that each species belongs to each main ecological category. This represents a paradigm shift and may change our vision of earthworm communities enabling the computation of the percentage of anecic, endogeic and epigeic species at the community level and thus overcoming the limits and debate about fixed ecological categories for each species.