The replacement of cis proline 39 of ribonuclease T1 by an alanine residue leads to a decrease in stability by about 20 kJ/mol and to major changes in the folding kinetics that are not easily explained by the proline model for protein folding. In particular, a novel very slow reaction is observed in the refolding of the Pro39Ala variant. Here the unfolding and refolding kinetics of this protein are further investigated. We show that the very slow reaction is not a prolyl isomerization. It is not created by a slow isomerization of the unfolded protein, nor is it catalyzed by prolyl isomerase, and all molecules have to undergo this reaction during refolding. Most of the unfolded Pro39Ala molecules contain an incorrect trans isomer at the remaining cis Pro55. They use a sequential pathway for refolding, in which trans to cis isomerization at Pro55 precedes the very slow reaction. The refolding of the minor fraction of unfolded Pro39Ala molecules with a correct cis isomer at proline 55 is a single first-order reaction that is limited in rate by the very slow step. The folding mechanism of wild-type ribonuclease T1 cannot be used to explain these results and independent mechanisms are proposed to model the unfolding and refolding of the Pro39Ala variant. The molecular interpretation of the changes in the folding mechanism is tied to the question, as to whether the cis character of the peptide bond at position 38-39 is maintained after the substitution of Pro39 by alanine. A possible explanation could be that the novel very slow folding reaction involves the trans to cis isomerization of the Tyr38-Ala39 bond. Such a reaction is probably slow, since the activation energy is high and since tight coupling with the formation of structure is required to stabilize the cis form of a non-prolyl peptide bond. Alternatively, the strong decrease in folding rate could be correlated with the general destabilization of ribonuclease T1 by the Pro39Ala mutation.