Reaction norms for age and size at maturity are being analyzed to answer important questions about the evolution of life histories. A new statistical method is developed in the framework of time-to-event data analysis, which circumvents shortcomings in currently available approaches. The method emphasizes the estimation of age-and size-dependent maturation rates. Individual probabilities of maturation during any given time interval follow by integrating maturation rate along the growth curve. The integration may be performed in different ways, over ages or sizes or both, corresponding to different assumptions on how individuals store the operational history of the maturation process. Data analysis amounts to fitting generalized nonlinear regression models to a maturation status variable. This technique has three main advantages over existing methods: (1) treating maturation as a stochastic process enables one to specify a rate of maturation; (2) age and size at which maturation occurs do not have to be observed exactly, and bias arising from approximations and interpolations is avoided; (3) ages at which sizes are measured and maturation status are observed can differ between individuals. An application to data on the springtail Folsomia candida is presented. Models with age-dependent integration of maturation rates were preferred. The analysis demonstrates a significant size dependence of the maturation rate but no age dependence.