The study of within-person change lies at the core of developmental research. Theory and empirical data suggest that many of these developmental processes are not linear. We describe a broad class of multilevel models that allows for nonlinear change — nonlinear mixed models. To demonstrate the utility of these models, we present a nonlinear mixed model analysis of adjustment to conjugal loss. Coming from a perspective of the individual as a regulatory system, our model predicts a faster rate of adjustment immediately following the loss and diminished adjustment as time since the loss increases, approaching an equilibrium level of well-being. This model allows us to estimate various aspects of the adjustment trajectory and individual differences in these trajectories, including multiple ways that pre- and post-loss factors can explain variability in the adjustment process. The model provides new insights into an important phenomenon that cannot be gleaned from linear models and other methods of trajectory analysis. We discuss the strengths and limitations of this type of analysis relative to other methods.