Background: Empathy is a core competency in undergraduate nursing education and is closely linked to quality of care. Its expression may be influenced by individual regulatory traits such as resilience, particularly in emotionally demanding academic clinical contexts. However, the dimensional associations between resilience and empathy remain insufficiently explored in nursing students. Objective: To examine structural associations between resilience and empathy dimensions among undergraduate nursing students from multiple Latin American countries. Methods: A cross-sectional multicenter study was conducted with 3611 undergraduate nursing students from five Latin American countries. Participants completed the Jefferson Scale of Empathy-Health Professions Student version (JSE-HPS) and the Resilience Trait Scale (RTS). Confirmatory factor analyses and reliability analyses were performed for both instruments. Measurement invariance across countries was evaluated for the resilience scale. Structural associations between resilience and empathy dimensions were estimated using structural equation modeling and interpreted as theoretically guided exploratory approximations. Results: The structural model showed inadequate incremental fit (CFI = 0.83; TLI = 0.81), despite comparatively more favorable RMSEA and SRMR values. Consequently, the individual path coefficients cannot be regarded as stable or well-supported structural associations. Within this misspecified model, the estimated coefficients suggested heterogeneous patterns across resilience and empathy dimensions; however, their magnitude, direction, and statistical significance may be sensitive to model specification. These results should therefore be considered preliminary and hypothesis-generating rather than as reliable estimates of dimensional relationships. Conclusions: The inadequate incremental fit of the structural model precludes firm conclusions regarding the dimensional associations between resilience and empathy. Although the estimated coefficients suggested potentially differentiated patterns, these findings are too model-dependent to support educational recommendations or substantive interpretations at this stage. Further studies should refine and cross-validate the measurement and structural specifications before evaluating the educational relevance of these relationships.