Cross-slip of screw dislocations is a dislocation process involved in dislocation structuring, work hardening, and fatigue. Cross-slip nucleation in FCC solid solution alloys has recently been shown to be strongly influenced by local fluctuations in spatial arrangement of solutes, leading to a statistical distribution of cross-slip nucleation barriers. For cross-slip to be effective macroscopically, however, small cross-slip nuclei (~40b) must expand across the entire length of typical dislocation segments (10 2-10 3 b). Here, a model is developed to compute the relevant activation energy distribution for cross-slip in a random FCC alloy over arbitrary lengths and under non-zero Escaig and Schmid stresses. The model considers cross-slip as a random walk of successive flips of adjacent1b segments, with each flip having an energy consisting of a deterministic contribution due to constriction formation and stress effects, plus a stochastic contribution. The corresponding distribution is computed analytically from solute-dislocation and solute-solute binding energies. At zero stress, the probability of high activation energies increases with dislocation length. However, at stresses of just a few MPa, these barriers are eliminated and lower barriers are dominant. For increasing segment length, the effective energy barrier decreases according to a weak-link scaling relationship and good analytic predictions can be made using only known material properties. Overall, these results show that the effective cross-slip barrier in a random alloy is significantly lower than estimates based on average elastic and stacking fault properties of the alloy.