The temperature dependence of the (1)H chemical shifts of six designed peptides previously shown to adopt beta-hairpin structures in aqueous solution has been analyzed in terms of two-state (beta-hairpin left arrow over right arrow coil) equilibrium. The stability of the beta-hairpins formed by these peptides, as derived from their T(m) (midpoint transition temperature) values, parallels in general their ability to adopt those structures as deduced from independent NMR parameters: NOEs, Deltadelta(C)(alpha)(H), Deltadelta(C)(alpha), and Deltadelta(C)(beta) values. The observed T(m) values are dependent on the particular position within the beta-hairpin that is probed, indicating that their folding to a beta-hairpin conformation deviates from a "true" two-state transition. To obtain individual T(m) values for each hairpin region in each peptide, a simplified model of a successive uncoupled two-state equilibrium covering the entire process has been applied. The distribution of T(m) values obtained for the different beta-hairpin regions (turn, strands, backbone, side chains) in the six analyzed peptides reveals a similar pattern. A model for beta-hairpin folding is proposed on the basis of this pattern and the reasonable assumption that regions showing higher T(m) values are the last ones to unfold and, presumably, the first to form. With this assumption, the analysis suggests that turn formation is the first event in beta-hairpin folding. This is consistent with previous results on the essential role of the turn sequence in beta-hairpin folding.