Purpose <p>This study aims to evaluate and validate land-use Life Cycle Impact Assessment (LCIA) methods for assessing ecosystem services (ES) and soil biodiversity (SB) in Brazilian feed production systems. Using field- and farm-scale empirical data, it examines the ability of these methods to capture spatial and temporal variability and differences between organic and conventional management, generating evidence from a major global food-producing region that remains underrepresented in Life Cycle Assessment studies.</p> Methods <p>The analysis focused on key indicators of ES and SB, including changes in soil organic carbon and bulk density, erosion potential, infiltration reduction potential, physicochemical filtration reduction potential, groundwater regeneration reduction potential, and the soil biodiversity threat index. Model validation combined a quantitative metric, the Index of Agreement (IA), with graphical analysis of uncertainty overlap to assess consistency between model predictions and observed field data. The inventory data comprise three systems: two large-scale organic feed systems, a comparable conventional feed-producing farm in the same region and under the same pedoclimatic conditions, and legally protected ‘natural sites’ located within the boundaries of the organic farms (reference state).</p> Results and discussion <p>The results indicate that the performance of LCIA models heavily depends on the availability of high-quality, locally relevant input data. Models that include site-specific information, especially for soil biodiversity and bulk density, effectively captured local variability and showed high agreement with observations (IA = 0.95 and 0.96). Models relying on generic datasets tend to underestimate impacts related to erosion, infiltration, groundwater recharge, and physicochemical filtration due to their oversimplified structures and lack of regional detail. Overall, organic systems outperform conventional ones in environmental performance, particularly regarding soil organic carbon and biodiversity, highlighting advantages such as reduced tillage and pesticide use. Nonetheless, current LCIA methods have difficulty accurately representing spatial and temporal variability, crop differences, and long-term soil processes, reducing their accuracy in tropical regions.</p> Conclusions <p>Although existing LCIA methods offer valuable insights into land-use sustainability in Brazil, further methodological improvements are necessary to boost their reliability. Incorporating region-specific parameters, site-level data, and the dynamic features of tropical agricultural systems, such as extreme weather events and diverse soil–crop interactions, is crucial for producing more accurate and policy-relevant sustainability assessments.</p>