Current fluorescence-based adhesion assays that use a 96-well plate format rely on the assumption that the fluorescent label does not significantly leak from the cells. Thus, we evaluated a calcein-based, in vitro adhesion assay in 96-well plates using five different types of leukocytes (HL60 cells, human neutrophils, rat neutrophils, mouse progenitor T cells and EL4 cells). Each cell type leaked calcein at a different rate, with the highest rates found for rat neutrophils and progenitor T cells, which lost as much as 20%-40% of the label within 90 min, the time required to complete the assay. Thus, we developed a procedure to measure the dye leakage rate during the assay in order to obtain a correction factor, which was then used to calculate the "true" number of adherent cells. Data for the adhesion of FTF1 cells to endothelial monolayers, after correction for calcein leakage, deviated less than 10% of adhesion data obtained with a well-established 51Cr-based assay.