Formation of bonds in thermally agitated environments is ubiquitous in biological, solid state, and nuclear physics with many engineering applications. Particles or molecular associations form binary complexes when there is an energy gain upon forming a bond. Examples include nanoparticle absorption to membranes, protein-ligand bindings, and atomic force microscopy (AFM) studies of the cellular membrane. The energy gain upon forming the bond, i.e. the binding energy Q, also determines the bond's stability in an environment with thermal fluctuations. When the binding energy is small compared to thermal fluctuations that have energies of the order kBT, i.e. when Q= kBT ~ 1, the bond is unstable and can break on a short time scale. Conversely, for large binding energies, Q >> kBT, the bond is stable, and the dissociation time is very long. Current theories are successful in the latter case of large binding energy, but fail completely in the opposite limit of low energy/large thermal fluctuations. The goal of this PhD is to fill this gap in our understanding of thermally-activated dissociation processes. Stochastic methods will be employed and combined creatively with Zwanzig-Caldeira-Leggett system-bath Hamiltonian methods to properly describe the role of friction. The theory will be applied to selected important problems such as biomolecular receptor-ligand binding, biological filament dissociation, and alike.