LHAPDF6 is a re-engineered library for accessing parton density functions (PDFs) in high-energy physics, addressing the limitations of its predecessor, LHAPDF5. It is designed for use in the LHC precision era, offering improved performance, scalability, and accuracy. The new library is built in C++ and provides a more flexible and efficient framework for handling PDF data, including support for multiple PDF sets, reduced memory usage, and enhanced metadata handling. Over 200 PDF sets have been migrated to the new format, ensuring compatibility with a wide range of Monte Carlo generators and physics programs. LHAPDF6 supports both central and error PDF sets, allowing for systematic variations and reweighting of PDFs. It introduces a cascading metadata system that enables efficient access to PDF information, including parameters like α_S and Λ_QCD. The library also includes advanced interpolation and extrapolation methods for accurate PDF value calculations. The design of LHAPDF6 addresses key issues in the previous version, such as memory management, correctness of metadata, and maintainability, making it suitable for the demanding requirements of LHC Run 2 and beyond. The library is available in C++, Python, and Fortran, with a Python interface for interactive PDF testing. Data formats for PDFs are standardized, with metadata stored in YAML files and PDF grid data in a structured format. The library also includes tools for managing PDF sets, including an index file for mapping LHAPDF IDs to PDF members and a script for updating PDF data. Overall, LHAPDF6 represents a significant improvement over its predecessor, providing a robust and efficient solution for PDF access in high-energy physics.LHAPDF6 is a re-engineered library for accessing parton density functions (PDFs) in high-energy physics, addressing the limitations of its predecessor, LHAPDF5. It is designed for use in the LHC precision era, offering improved performance, scalability, and accuracy. The new library is built in C++ and provides a more flexible and efficient framework for handling PDF data, including support for multiple PDF sets, reduced memory usage, and enhanced metadata handling. Over 200 PDF sets have been migrated to the new format, ensuring compatibility with a wide range of Monte Carlo generators and physics programs. LHAPDF6 supports both central and error PDF sets, allowing for systematic variations and reweighting of PDFs. It introduces a cascading metadata system that enables efficient access to PDF information, including parameters like α_S and Λ_QCD. The library also includes advanced interpolation and extrapolation methods for accurate PDF value calculations. The design of LHAPDF6 addresses key issues in the previous version, such as memory management, correctness of metadata, and maintainability, making it suitable for the demanding requirements of LHC Run 2 and beyond. The library is available in C++, Python, and Fortran, with a Python interface for interactive PDF testing. Data formats for PDFs are standardized, with metadata stored in YAML files and PDF grid data in a structured format. The library also includes tools for managing PDF sets, including an index file for mapping LHAPDF IDs to PDF members and a script for updating PDF data. Overall, LHAPDF6 represents a significant improvement over its predecessor, providing a robust and efficient solution for PDF access in high-energy physics.