We use a spatially explicit landscape model to investigate the potential role of rainfall on shrub-grass transitions in the Jornada Basin of southern New Mexico during the past century. In long-term simulations (1915 - 1998) along a 2700m transect running from a dry lake bed to the foothills of a small mountain, we test two hypotheses: (i) that wetter winters and drier summers may have facilitated shrub encroachment in grasslands, and (ii) that increases in large precipitation events may have increased soil water recharge at deeper layers, thus favoring shrub establishment and growth. Our model simulations generally support the hypothesis that wetter winters and drier summers may have played a key role, but we are unable to reproduce the major shifts from grass- to shrubdomination that occurred in this landscape during the early part of the 1900s; furthermore, the positive shrub response to wetter winters and drier summers was only realized subsequent to the drought of 1951 - 1956, which was a relatively short ‘window of opportunity’ for increased shrub establishment and growth. Our simulations also generally support the hypothesis that an increase in the number of large precipitation events may also have favored shrub establishment and growth, although these results are equivocal, depending upon what constitutes a ‘large’ event and the timing of such events. We found complex interactions among (i) the amount/seasonality of rainfall, (ii) its redistribution in the landscape via run-on and runoff, (iii) the depth of the soil water recharge, and (iv) subsequent water availability for the growth and reproduction of shrubs vs. herbaceous plants at various landscape positions. Our results suggest that only a mechanistic understanding of these interactions, plus the role of domestic cattle grazing, will enable us to elucidate fully the relative importance of biotic vs. abiotic factors in vegetation dynamics in this semiarid landscape.