The lambda(6-85)(*) pseudo-wild type of lambda repressor fragment is a fast two-state folder (k(f) approximately 35 microsec(-1) at 58 degrees C). Previously, highly stable lambda(6-85)(*) mutants with k(f) > 30 microsec(-1) have been engineered to fold nearly or fully downhill. Stabilization of the native state by solvent tuning might also tune lambda(6-85)(*) away from two-state folding. We test this prediction by examining the folding thermodynamics and kinetics of lambda(6-85)(*) in a stabilizing solvent, 45% by weight aqueous ethylene glycol at -28 degrees C. Detection of kinetics by circular dichroism at 222 nm (sensitive to helix content) and small angle X-ray scattering (measuring the radius of gyration) shows that refolding from guanidine hydrochloride denatured conditions exhibits very different time scales for collapse and secondary structure formation: the two processes become decoupled. Collapse remains a low-barrier activated process, while the fastest of several secondary structure formation time scales approaches the downhill folding limit. Two-state folding of lambda(6-85)(*) is not a robust process.