The 2nd EARL Workshop on Evaluating and Applying Recommender Systems with Large Language Models will take place at RecSys 2025 in Prague, Czech Republic. This workshop focuses on advancing the application, evaluation, and understanding of large language models (LLMs) in the context of recommender systems (RSs). The event encourages submissions on emerging techniques such as retrieval-augmented generation (RAG), multi-modal recommendation, reinforcement learning with human feedback (RLHF), dynamic prompting, scalable fine-tuning, and personalized conversational agents. It also highlights real-world applications, deployment experiences, and critical challenges related to fairness, transparency, and responsible AI in LLM-driven RSs.