The role played by entropy for the enormous rate enhancement achieved by enzymes has been debated for many decades. There are, for example, several confirmed cases where the activation free energy is reduced by around 10 kcal/mol due to entropic effects, corresponding to a rate enhancement of ∼107 compared to the uncatalyzed reaction. However, despite substantial efforts from both the experimental and theoretical side, no real consensus has been reached regarding the origin of such large entropic contributions to enzyme catalysis. Another remarkable instance of entropic effects is found in enzymes that are adapted by evolution to work at low temperatures, near the freezing point of water. These cold-adapted enzymes invariably show a more negative entropy and a lower enthalpy of activation than their mesophilic orthologs, which counteracts the exponential damping of reaction rates at lower temperature. The structural origin of this universal phenomenon has, however, remained elusive. The basic problem with connecting macroscopic thermodynamic quantities, such as activation entropy and enthalpy derived from Arrhenius plots, to the 3D protein structure is that the underlying detailed (microscopic) energetics is essentially inaccessible to experiment. Moreover, attempts to calculate entropy contributions by computer simulations have mostly focused only on substrate entropies, which do not provide the full picture. We have recently devised a new approach for accessing thermodynamic activation parameters of both enzyme and solution reactions from computer simulations, which turns out to be very successful. This method is analogous to the experimental Arrhenius plots and directly evaluates the temperature dependence of calculated reaction free energy profiles. Hence, by extensive molecular dynamics simulations and calculations of up to thousands of independent free energy profiles, we are able to extract activation parameters with sufficient precision for making direct comparisons to experiment. We show here that the agreement with the measured quantities, for both enzyme catalyzed and spontaneous solution reactions, is quite remarkable. Importantly, we can now address some of the most spectacular entropy effects in enzymes and clarify their detailed microscopic origin. Herein, we discuss as examples the conversion of cytidine to uridine catalyzed by cytidine deaminase and reactions taking place on the ribosome, namely, peptide bond formation and GTP hydrolysis by elongation factor Tu. It turns out that the large entropy contributions to catalysis in these cases can now be rationalized by our computational approach. Finally, we address the problem of cold adaptation of enzyme reaction rates and prove by computational experiments that the universal activation enthalpy-entropy phenomenon originates from mechanical properties of the outer protein surface.