Measurement of serum folate levels in humans using a radioassay has now been frequenly reported. However, very little data is available assessing the diagnostic value of this test in comparison with other laboratory and clinical information pertaining to folate status. A radioassay, using the folate binder in dried milk was used in parallel with the well established Lactobacillus casei microbiological assay. Three slightly different modifications of the radioassay were tested. At first, a partially purified milk binder was used, in the second the unpurified milk and, third, the unpurified milk with variation of the incubation periods during the assay, were utilized. A total of 159 sera were tested by one of the modifications of the radioassay as well as by the L. casei method. Duplicate estimations were done in all cases. Using the first method, the correlation between the radio- and microbiological assays was 0.89, with the second 0.52, and with the third 0.85. The final assay system gave the most reliable results when compared to the L. casei assay which was taken as the standard of reference. However, in a number of cases, normal levels by L. casei gave low levels by radioassay and vice versa. In 13 such instances, clinical and other laboratory data were assessed and it was concluded that the microbiological assay gave the more accurate indication of folate stores in humans. Some reasons for the discrepancies between the assay systems were discussed.