We present a technique for estimating size-age relations and size-dependent mortality patterns of long-lived plants. The technique requires two sets of size data of individual (non-marked) plants that should be collected with a time-lag of several years in the same area of a study site. The basic idea of our technique is to assume general (three parameter) families of size-dependent functions which describe growth and mortality that occurred between the two data gathering events. We apply these growth and mortality functions to the size data of the early data set and construct predicted size-class distributions to compare it, in a systematic way, to the size-class distribution of the later data set. In a next step we calculate the size-age relations from the resulting growth functions, which yield the smallest difference between observed and predicted size-class distribution. Applying this technique to size data of five dominant shrub species at the Tierberg study site in the semiarid Karoo, South Africa produced new insight into the biology of these species which otherwise cannot be obtained without frequent measurements of marked plants. We could relate characteristics of growth behavior and mortality, for certain subgroups of the five species, to the life-history attributes evergreen vs. deciduous, succulent vs. woody, and early reproductive vs. late reproductive. The results of our pilot-study suggest a broad applicability of our technique to other shrublands of the world. This requires at least one older record of (individual) shrub-size data and performance of resampling.