The prolylproline sequence unit is found in several naturally occurring linear and cyclic peptides with immunosuppressive and toxic activity. Furthermore, Pro–Pro units are abundant in collagen, in ligand motifs binding to SH3 or WW domains, as well as in vital enzymes such as DNA glycosylase and thrombin. In all these sequence units a special role is dedicated to conformation in order to successfully fulfill the appropriate biological function. Therefore, a detailed analysis of the basic conformational properties of Pro–Pro is expected to reveal the versatile structural role of this sequence. PCM (polarizable continuum model) calculations on the basis of ab initio and density functional theory investigations using the model peptide HCO–L-Pro–L-Pro–NH2 are presented. Cis–trans isomerism, backbone conformation and ring puckering are studied. A systematic comparison is made to experimental data gained on L-prolyl–L-proline sequence units retrieved from the Protein Data Bank as well as from the Cambridge Structural Database. PCM data are in good agreement with high-resolution X-ray crystallography. Population data derived from energy calculations and those gained directly from statistics predict that 87% of the Pro–Pro sequence units adopt elongated structures, while 13% form β-turns. Both approaches prefer the same 6 out of the 36 ideally possible backbone folds. Polyproline II unit (tϵLtϵL), other elongated structures (cϵL tϵL, tϵL tαL and tϵL tγL), type VIa (tϵL cαL) and type I or III β-turns (tαL tαL) altogether describe 96% of the prolylproline sequences. In disordered proteins or domains, Pro–Pro sequence units may sample the various conformers and contribute to the segmental motions. Proteins 2008. © 2007 Wiley-Liss, Inc.