From the early 1980s to the early 1990s, the incidence of cancer in American children under 10 years of age rose 37 percent, or 3 percent annually. There is an inverse correlation between increases in cancer rates and age at diagnosis; the largest rise (54 percent) occurred in children diagnosed before their first birthday. Rates rose for all 11 states and cities included in the analysis. A jump in cancer rates for children born in 1982-83 was followed by a drop; but another abrupt rise for the 1986-87 birth cohort has been sustained thereafter. Results indicate that the rising childhood cancer rate represents a far more serious problem in the United States than previous reports have suggested. The methodology used here adds three additional states and cities, analyzes children under 10 rather than under 15, begins the analysis in 1980 rather than in 1973, and extends the study to 1993, which may partially account for the new findings. There are no apparent explanations for these trends, suggesting that researchers should analyze the data more fully and propose hypotheses on potential causes. One possible factor, fetal and infant exposure to low-dose radioactivity, is explored here.