Our objective was to determine if long-term increases in precipitation can maintain grasslands susceptible to desertification, and initiate a reversal of historic regime shifts on desertified shrublands. Perennial grass production and species richness in a multi-year wet period were hypothesized to be greater than expected based on precipitation in a sequence of dry years. These responses were expected to differ for grasslands and shrublands with different dominant species and topo-edaphic properties. Long-term trends in desertification were documented using vegetation maps beginning in 1858, 1915, 1928, and 1998). These trends were compared with herbaceous and woody species responses to a sequence of dry (1994–2003) and wet years (2004–2008) for two grassland (uplands, playas) and three desertified shrubland types (honey mesquite, creosotebush, tarbush) in the Chihuahuan Desert. Analyses showed that both types of grasslands decreased in spatial extent since 1858 whereas areas dominated by mesquite or creosotebush increased. Production of upland grasslands in the wet period was greater than expected based on responses during the dry period whereas the relationships between species richness and precipitation was the same for both periods. Precipitation was not important to responses in playa grasslands in either period. For all ecosystem types, the production response in wet years primarily was an increase in herbaceous plants, and the most pronounced responses occurred on sandy sites (upland grasslands, mesquite shrubland). Results suggest that multiple wet years are needed to initiate a sequence of grass establishment and survival processes that can maintain upland grasslands without management inputs and lead to a state change reversal in desertified shrublands. Restoration strategies need to take advantage of opportunities provided by future climates while recognizing the importance of ecosystem type.