<p>The 2023–24 Nigeria Demographic and Health Survey indicates that child malnutrition remains a major public health challenge, with approximately 40% of children under five being stunted, 8% wasted, and 27% underweight. Anthropometric indicators such as stunting, wasting, and underweight are widely used to assess child nutritional status. While previous studies have identified individual and community-level determinants, less is known about how these factors operate across hierarchical levels and how they vary spatially. This study applies a multilevel structured additive regression (STAR) model within a Bayesian framework to examine childhood malnutrition across child, community, and state levels. Data were obtained from the 2023–2024 Nigeria Demographic and Health Survey and the 2023 National Bureau of Statistics datasets. Model estimation was carried out using Markov chain Monte Carlo methods with appropriate prior specifications for all parameters. Variance decomposition shows that a substantial proportion of unexplained variation is attributable to differences between states, followed by community-level effects. Results indicate that spatially structured effects, state-level unemployment, child age, and maternal age are important predictors of nutritional outcomes. The spatial effects reveal localized clustering of higher levels of malnutrition in parts of the North-East, with additional smaller clusters observed in some southern states. At the same time, many states across both northern and southern regions show relatively lower or average levels, suggesting substantial within-country heterogeneity rather than a clear national gradient. These findings highlight the importance of multilevel and spatial modelling approaches in understanding the drivers of under-five malnutrition and provide evidence to support geographically targeted interventions aimed at reducing nutritional disparities in Nigeria.</p>