Ingestion of soil by young children may be an important source of exposure to environmental contaminants. Estimates of soil ingestion have been made by several studies using trace elements in a mass-balance approach, but differ substantially between studies and trace elements. We conduct a meta-analysis of four major mass-balance soil ingestion studies conducted on children between one and seven in the summer/fall in the northern United States. The analysis takes advantage of primary data from all studies, and provides a more complete description of soil ingestion among children. The meta-analysis uses data based on the two most reliable trace elements, Al and Si, that have passed a screening to identify and exclude measures with a high likelihood of bias. Details are described in a companion paper. The best linear unbiased predictor is used in a mixed model to estimate soil ingestion for study subjects. Overall, 11% of subject-periods are identified as outliers and excluded from the analysis. An analysis on 216 children based on Al and Si as tracer elements indicates that the mean (median) soil ingestion is 26 mg/day (33 mg/day), with the 95th percentile estimated as 79 mg/day. This systematic approach provides more reliable estimates than individual study results. There is some evidence that soil ingestion increases with a child's age, but insufficient data to distinguish soil ingestion by gender.