Kinetics of folding of a protein held in a force-clamp are compared to an unconstrained folding. The comparison is made within a simple topology-based dynamical model of ubiquitin. We demonstrate that the experimentally observed variations in the end-to-end distance reflect microscopic events during folding. However, the folding scenarios in and out of the force-clamp are distinct.Recent advances in nano-technology have enabled manipulation of single biomolecules, especially by means of the atomic force microscopy (AFM). The manipulation usually involves mechanical stretching and monitoring of the force of resistance as a function of displacement of the AFM tip. In 2001, Oberhauser et al.1 have developed a force-clamp -a variant of AFM with an electronic adjustment of the tip displacement so that a constant pulling force is maintained. This technique allows one to measure the force dependence of the unfolding probability and has been used to probe the mechanical stability of two different domains of titin. The force-clamp microscopy has been subsequently developed by Fernandez and Li 2 to monitor the folding trajectory of a single protein that is first stretched by a constant unfolding force and then suddenly submitted to a substantially reduced force. The first tests on polyubiquitin have demonstrated a structured time dependence of the end-to-end distance, L. What this behavior corresponds to microscopically remains to be elucidated.In this paper we ask what one can learn from monitoring L during folding of a protein under a small stress and, in particular, is this process related to folding that is taking place without any mechanical constraints? We address these questions theoretically by performing molecular dynamics simulations in a coarse-grained topology-based model 4 . Such a model is ideally suited to study conceptual questions about large conformational changes because it makes relevant time scales accessible to computations.We focus on ubiquitin and two-ubiquitin since this case relates to the experimental studies 2,3 . A single ubiquitin consists of 76 amino acids and its structure is deposited in the Protein Data Bank 5 with a code 1ubq. The two-ubiquitin is modeled by linking two ubiquitins in a series through an extra peptide bond. We follow the implementation of the model along the lines outlined in refs.6 .In short, a protein is represented by a chain of C α atoms that are tethered by harmonic potentials with minima at 3.8Å. The effective self-interactions between the atoms are either purely repulsive or are minimum-endowed-contacts of the Lennard-Jones type, V ij = 4ǫ . The length parameters σ ij are chosen so that the potential minima correspond, pair-by-pair, to the experimentally established native distances between the C α atoms in amino acids in the pair. The distinction between the two kinds of the interactions is established based on the absence or presence of overlaps between all other atoms in the i and j amino acids in the native conformation. The effective geometry of atoms in the t...