This study analyzes structural forces affecting state patterns of parental presence within grandparent-grandchild coresidence by testing demographic, social change, policy environment, and social problems models. The project combines published state-level data with the 1970, 1980, and 1990 Census Public Use Microdata Samples. While factors differentially affect the proportion of children living in each household structure, coresidence seems to be encouraged by social contexts in which people are poor, the middle generation is likely to die, and which have higher rates of births to teens. State policy environment only affects proportions of children living in three generation households for 1980 and 1990. Structural variables other than social problems matter more for percent of children living in three generation households than for proportion of children living only with grandparents. Structural variables explain more variation in proportion of children living in skipped generation households than in proportion of children living with both parents and grandparents. (Contains 5 tables.)