This paper employs 5 percent samples of the Guatemalan censuses of 1964 and 1973 to estimate differential childhood mortality. For 1973, conventional indirect mortality estimation procedures were applied; for 1964, a modification of the Preston-Palloni technique was used. A major result is that differential child mortality widened appreciably between about 1959 and about 1968. Factors leading to mortality decline were possibly unequally spread by geographic region, rural-urban residence, ethnic groups, and educational groups. Guatemala City seems to have benefited most from the decline. Analysis of small geographic units supports these conclusions and suggests some role for public health. A by-product of the analysis was the finding that parity was understated in the 1964 census.