Ex-vivo whole blood assays have been evaluated for their ability to accurately predict the risk of a first-dose cytokine reaction developing in vivo following therapeutic antibody infusion. Tumour necrosis factor alpha (TNF alpha) release was rapidly detected in cultures incubated with either anti-CD52 antibodies of the human IgG1 or rat IgG2b isotype, and to a lesser extent with a human IgG4 isotype. Endotoxin contamination of the antibodies was not responsive for cytokine release, since polymixin B failed to inhibit cytokine release using concentrations of this antibiotic which neutralized the enhanced cytokine release seen from LPS-spiked antibody. A rat IgG2b antibody to CD45 and a human IgG1 anti-CD3 also induced significant TNF release, however, an aglycosyl anti-CD3 mutant devoid of adverse side-effects in vivo, did not result in cytokine release in vitro. Since the pattern of cytokine release seen following the clinical use of these antibodies was in good agreement with the findings of the ex-vivo whole cultures, this demonstrates the usefulness of this assay to predict cytokine release in vivo.