Friction force microscopy and single-molecule force spectroscopy are experimental methods to explore multistable energy landscapes by means of a controlled reduction of the energy barriers between adjacent potential minima. This affects the system's interstate transition rates proportional to e−ΔE(f)/kBT, with ΔE(f) being the barrier height, kBT the thermal energy, and f the elastic force applied. It is often assumed that, at large forces, the barrier height scales as (fc − f)3/2, where fc is the critical force, at which the barrier vanishes. We show that, for the elastic forces produced by a pulling device of finite stiffness κ, this scaling relation is actually incorrect. Rather, the barrier is a double-valued function of force of the form , where f0 is the maximal force that the system potential can generate, and the characteristic stiffness κc is not necessarily much larger than κ. In particular, for finite κ, the barrier vanishes at a certain force fκ < f0, but, in view of the double-valuedness of ΔE(f), the maximal force f0 can still be reached. We derive the relation between the most probable force at the moment of transition, fm, and the pulling velocity, v. The usually assumed scaling fm ∝ (ln v)2/3 is recovered as the κ → 0 limit of our more general result, but becomes increasingly worse as κ grows. We introduce a new data analysis method that allows one to quantitatively characterize the system potential and evaluate the stiffness of the pulling device, κ, which is usually not known beforehand. We demonstrate the feasibility of our method by analyzing the results of a numerical experiment based on the standard Prandtl–Tomlinson model of nanoscale friction.