We studied the survival of a population-based cohort of 154 infants with spina bifida who were born during the eight-year period from 1972 through 1979 to residents of Atlanta. Our objectives were to describe the cohort's survival experience and to explore relationships between clinical characteristics and survival. Overall, 57% of the cohort survived one year or more. This figure is misleading, however, because it conceals important differences in survival among subgroups of affected infants. From univariate analyses, we observed significant differences in survival among infants categorized by year of birth, birth weight, the open-closed status of the defect, the highest level of the defect on the spine, the presence of multiple major birth defects, and the presence of hydrocephalus at birth. More infants born in the late 1970s survived their first year of life than infants born in the early 1970s; infants with open defects had lower survival than those with closed defects; and infants whose defects were low on the spine had better survival than those whose defects were higher. When comparing the survival experience of this cohort with that of other groups from other areas or from more recent years, health workers must consider referral biases and differences in the distribution of clinical characteristics.