The first two years' experience with the substitution transfusion in the Los Angeles area is reported and analyzed. Forty-five cases were substituted. Results of therapy are compared with 35 cases studied at the Los Angeles Children's Hospital during the same period but which were not treated with a substitution transfusion. Mortality figures favored the use of the conservative, multiple simple transfusion method of treatment in the series reported. Even in groups of comparable severity, the substituted cases showed a higher mortality. Evidence is presented that errors in technic may have had an appreciable effect on the poor showing made by the substitution transfusion method of therapy. However, such errors may be inherent in any radical procedure. The series of cases is analysed as to mortality, development of kernicterus, effect of previous pregnancies, effect of previous siblings with erythroblastosis, effect of transfusions, correlation of the Coomb's test with severity of disease, correlation of maternal antibody titer with severity of disease, transfusions needed for a cure and length of stay in the hospital. An objective standard to measure the severity of an individual case is suggested. Although empiric, it allows for summations and resolves a case down to a numerical figure. Thus the case can be compared to other cases in an analysis of any phase of the disease. Use of the Coomb's (or antiglobulin) test is described. This test was found to be reliable, liable, easy to perform and extremely sensitive. There was a good correlation between strength of the test and severity of a case. Substitution transfusions were not performed unless the test was positive. Errors in performing the substitution transfusion are described. These errors included using too little blood, delaying the operation too long, giving blood into the umbilical vein, injecting calcium improperly and failing to protect the infant against exposure and dehydration. A technic for the substitution transfusion is presented in detail with suggestions for avoiding the errors made in these cases. The theoretic advantages inherent in the substitution transfusion are enumerated.