The paper introduces GADGET, a code designed for cosmological simulations of structure formation and interacting galaxies. GADGET employs both the N-body approach for self-gravitating collisionless fluids and smoothed particle hydrodynamics (SPH) for collisional gas. The code supports periodic boundary conditions using Ewald summation and allows individual and adaptive timesteps for all particles. It uses a tree algorithm to compute gravitational forces, with an option to use special-purpose hardware like GRAPE. GADGET has been successfully applied to simulations with up to \(7.5 \times 10^7\) particles, including studies of large-scale structure formation, high-resolution cluster simulations, and interacting galaxies. The paper details the numerical algorithms used, including the gravitational force computation, SPH implementation, and time integration scheme, and provides tests of the code. Both the serial and massively parallel versions of GADGET are publicly released.The paper introduces GADGET, a code designed for cosmological simulations of structure formation and interacting galaxies. GADGET employs both the N-body approach for self-gravitating collisionless fluids and smoothed particle hydrodynamics (SPH) for collisional gas. The code supports periodic boundary conditions using Ewald summation and allows individual and adaptive timesteps for all particles. It uses a tree algorithm to compute gravitational forces, with an option to use special-purpose hardware like GRAPE. GADGET has been successfully applied to simulations with up to \(7.5 \times 10^7\) particles, including studies of large-scale structure formation, high-resolution cluster simulations, and interacting galaxies. The paper details the numerical algorithms used, including the gravitational force computation, SPH implementation, and time integration scheme, and provides tests of the code. Both the serial and massively parallel versions of GADGET are publicly released.