A recently developed radioimmunoassay for PAF was applied to blood extracts with the aims of defining and overcoming problems that lead to erroneous results and establishing optimum conditions for the accurate determination of PAF levels. The high lipid content of blood was found to interfere with the assay, and it appeared that phosphatidylcholine and/or sphingomyelin might be among the lipids responsible. Interference was eliminated by either dilution or preparative TLC of blood extract prior to RIA although dilution is unlikely to be generally useful due to the low amounts of PAF normally present in human blood. Lipid extraction of whole blood followed by preparative TLC proved to be necessary in the preparation of samples prior to performance of the RIA. The problems encountered in the measurement of PAF levels in blood by RIA highlight the importance of determining the correct method of sample preparation for any tissue/fluid prior to its inclusion in the assay.