The kinetics of the helix<==>coil transition of an alanine-based peptide following a laser-induced temperature jump were monitored by the fluorescence of an N-terminal probe, 4-(methylamino)benzoic acid (MABA). This probe forms a peptide hydrogen bond to the helix backbone, which changes its fluorescence quantum yield. The MABA fluorescence intensity decreases in a single exponential relaxation, with relaxation times that are weakly temperature dependent, exhibiting a maximum value of approximately 20 ns near the midpoint of the melting transition. We have developed a new model, the kinetic version of the equilibrium 'zipper' model for helix<==>coil transitions to explain these results. In this 'kinetic zipper' model, an enormous reduction in the number of possible species results from the assumption that each molecule contains either no helical residues or a single contiguous region of helix (the single-sequence approximation). The decay of the fraction of N-terminal residues that are helical, calculated from numerical solutions of the kinetic equations which describe the model, can be approximately described by two exponential relaxations having comparable amplitudes. The shorter relaxation time results from rapid unzipping (and zipping) of the helix ends in response to the temperature jump, while the longer relaxation time results from equilibration of helix-containing and non-helix-containing structures by passage over the nucleation free energy barrier. The decay of the average helix content is dominated by the slower process. The model therefore explains the experimental observation that relaxation for the N-terminal fluorescent probe is approximately 8-fold faster than that for the infrared probe of Williams et al. [(1996) Biochemistry 35, 691-697], which measures the average helix content, but does not account for the absence of observable amplitude for the slow relaxation in the fluorescence experiments (<10% slow phase). If we assume that the activation barrier for the coil-->helix rate is purely entropic, the model can also explain the maximum in the temperature dependence of the relaxation time for the fluorescent probe. Parameters that best reproduce the melting curves and the ratio of relaxation times predict a value of the cooperativity parameter sigma which is approximately 3-fold larger than previously reported values obtained from fitting equilibrium data only. The helix growth rate of approximately 10(8) s-1 that reproduces the experimental relaxation times is approximately 100-fold slower than those observed in molecular dynamics simulations. These parameters can be used to simulate the kinetically cooperative formation of a helix from the all-coil state.