“All models are wrong but some are useful” (G. Box). In this presentation, I aim to show that research questions such as assessing the risk associated with wireworm infestation and the potential damage to crops can benefit from statistical and mathematical modelling. A wide variety of modelling approaches can be used. The first use, certainly the most traditional, relies on regression models based on data that can reveal relationships between a response variable such as wireworm abundance (or occurrence) or crop damage and potential explanatory descriptors. Among the many studies that apply this approach, I will discuss the analysis of long-term survey data on maize fields in France, where we examined the relative influence of putative key explanatory variables on wireworm damage and derived a model for predicting damage risk [Poggi et al, 2018]1. Depending on its generalisation capacity, such a model may form the cornerstone of a decision support system for the management of wireworm damage in maize crops. Beyond correlative approaches, certain modelling frameworks allow latent variables to be considered and inferred, which is particularly relevant when dealing with quantities that are difficult - if not impossible - to access, such as populations of belowground pests. Hierarchical Bayesian modelling offers this possibility and also provides an appropriate framework for dealing with risk assessment, since the model results are expressed as probabilistic distributions (called posteriors). By way of illustration, I will present an original hierarchical Bayesian model that explicitly considers biological knowledge (including three biological processes: mortality, oviposition and vertical migration) and the uncertainty of field observations (stochastic observation model), rather than relying solely on statistical correlations, to predict the level of wireworm infestation [Roche et al., 2023]2. Overall, the development of models that describe the mechanisms driving wireworm colonization, and subsequently elucidate the ecological processes operating at the landscape scale, remains an avenue for future research. In an initial attempt, we proposed a framework combining (i) a spatially explicit mechanistic model describing the population dynamics of the pest in the aerial and soil compartments involved throughout its life cycle, and (ii) spatio-temporal representations of various landscape contexts [Poggi et al, 2021]3. Once parameterised, in particular on the basis of the knowledge available in the scientific literature, this model opens the way to exploring the spatio-temporal manipulation of land use (e.g. the arrangement of grassy landscape elements) for pest management. All these examples are intended to contribute to the discussion on how to better assess the risks associated with wireworm infestation and the potential damage it can cause to crops, either by enriching the data to be analysed or by improving the integration of knowledge in order to better describe the mechanisms driving the wireworm infestation at plot or landscape scale.