AbstractThe solution structure of an α‐amylase inhibitor, tendamistat, calculated from nmr data with the distance geometry program DISMAN is subjected to restrained energy minimization. To study the influence of force field parametrizations and the convergence behavior of refinement algorithms, two different programs were used. AMBER is an established software package including a steepest descent and/or conjugent gradient optimizer in the Euclidian space; the name AMBER also represents a force field. The program FANTOM (fast Newton–Raphson torsion angle energy minimizer) is a new restrained energy refinement implementation of the Newton–Raphson algorithm, which uses second derivatives of the conformational energy in dihedral angle space with the ECEPP/2 force field. For both programs the normal energy force field was supplemented with an additional potential of the form ΣA(di−ui)6(ifdi>ui), which enforces upper limitsuito selected distancesdias measured by nmr. Improvements of the intramolecular interactions with a decrease of the internal energies of about 1000 kcal/mol could be achieved without increasing the distance constraint violations. The restrained energy refinements caused only small changes of the molecular geometries: The root mean square distance values for the backbone atoms between the initial DISMAN structure and the refined structures are about 0.5 Å for AMBER and about 0.7 Å for FANTOM. Local conformational changes during the restrained energy minimizations are analyzed with respect to hydrogen‐bond formation, and with respect to comparisons of the solution structure and the crys