LHAPDF6 is a redesigned and re-implemented version of the LHAPDF library, addressing the limitations of its Fortran predecessor. The original LHAPDF, developed to replace PDFLIB, faced issues with static memory allocation, limited concurrency, and inefficiencies in handling multiple PDF sets. These problems became critical during the LHC Run 1, prompting the development of LHAPDF6 to meet the demands of LHC Run 2 and beyond. LHAPDF6 is built in C++ and introduces a cascading metadata system, allowing for efficient and flexible access to parton density functions (PDFs). It supports concurrent PDF sets, reduces static memory usage, and improves CPU performance. The library uses a unified data format for all PDFs, with over 200 PDF sets migrated from LHAPDF5 through a rigorous quality control process. It is compatible with many Monte Carlo generators and other physics programs, including through full compatibility routines. The design of LHAPDF6 includes a powerful metadata system that allows for the storage and retrieval of information such as α_S values, parton flavours, and other parameters. This system enables efficient handling of PDF uncertainties and reweighting, crucial for precision physics studies. The library also features a flexible interpolator system, which supports various interpolation schemes, including cubic Hermite splines and linear interpolation, and an extrapolator system for handling values outside the grid range. LHAPDF6 addresses the limitations of the Fortran version by allowing dynamic memory management, supporting multiple PDF sets simultaneously, and providing a more robust and scalable framework. It also includes a Python interface, making it accessible for interactive PDF testing and exploration. The library is designed to be compatible with a wide range of PDF sets, including those with nuclear correction factors, and supports various PDF data formats, ensuring flexibility and efficiency in handling parton density functions for high-energy physics applications.LHAPDF6 is a redesigned and re-implemented version of the LHAPDF library, addressing the limitations of its Fortran predecessor. The original LHAPDF, developed to replace PDFLIB, faced issues with static memory allocation, limited concurrency, and inefficiencies in handling multiple PDF sets. These problems became critical during the LHC Run 1, prompting the development of LHAPDF6 to meet the demands of LHC Run 2 and beyond. LHAPDF6 is built in C++ and introduces a cascading metadata system, allowing for efficient and flexible access to parton density functions (PDFs). It supports concurrent PDF sets, reduces static memory usage, and improves CPU performance. The library uses a unified data format for all PDFs, with over 200 PDF sets migrated from LHAPDF5 through a rigorous quality control process. It is compatible with many Monte Carlo generators and other physics programs, including through full compatibility routines. The design of LHAPDF6 includes a powerful metadata system that allows for the storage and retrieval of information such as α_S values, parton flavours, and other parameters. This system enables efficient handling of PDF uncertainties and reweighting, crucial for precision physics studies. The library also features a flexible interpolator system, which supports various interpolation schemes, including cubic Hermite splines and linear interpolation, and an extrapolator system for handling values outside the grid range. LHAPDF6 addresses the limitations of the Fortran version by allowing dynamic memory management, supporting multiple PDF sets simultaneously, and providing a more robust and scalable framework. It also includes a Python interface, making it accessible for interactive PDF testing and exploration. The library is designed to be compatible with a wide range of PDF sets, including those with nuclear correction factors, and supports various PDF data formats, ensuring flexibility and efficiency in handling parton density functions for high-energy physics applications.