Residual dipolar couplings (RDCs) have been observed in disordered states of several proteins. While their nonuniform values were initially surprising, it has been shown that reasonable approximation of experimental RDCs can be obtained using simple statistical coil models and assuming global alignment of each structure, provided that many thousands of conformers are averaged. Here we show that, by using short local alignment tensors, we can achieve good agreement between experimental and simulated RDCs with far fewer structures than required when using global alignment. This makes the possibility of using RDCs as direct restraints in structural calculations of disordered proteins much more feasible. In addition, it provides insight into the nature of RDCs in disordered states, suggesting that they are primarily reporting on local structure.