Two colour flow cytometry was used to analyse in situ cytokine expression by human monocytes. Whole blood was cultured in siliconised glass bottles, with or without E. coli lipopolysaccharide (LPS), for various times, and the mononuclear cells (MNCs) then exposed to a variety of permeabilisation procedures prior to flow cytometric analysis. Paraformaldehyde (PF)/saponin fixation preserved cellular morphology, and caused a reproducible degree of permeabilisation (estimated by propidium iodide inclusion: mean 94%, range 86-99% (n = 33)). After fixation with 4% PF and permeabilisation with 1% saponin at 0 degrees C in PBS containing 20% human serum, MNCs were incubated with phycoerythrin(PE)-conjugated mouse anti-CD14 (monocyte phenotype) and polyclonal rabbit anti-human interleukin-1 alpha (IL-1 alpha), IL-1 beta, tumour necrosis factor alpha (TNF-alpha), or control rabbit IgG. Binding of rabbit antibodies was detected using goat anti-rabbit IgG fluorescein isothiocyanate (FITC). FITC fluorescence was increased in CD14 PE positive cells with the three anti-cytokine antibodies following LPS stimulation, compared with controls. There was a reproducible dose related response in monocyte IL-1 beta and TNF-alpha expression following LPS stimulation, with early peaks in TNF-alpha (2 h), compared with IL-1 beta (4 h), and IL-1 alpha (12 h). Specificity of this cytokine detection system was confirmed by inhibition studies using the corresponding recombinant human cytokines, by an absence of staining in CD14 negative or unpermeabilised MNCs, and by the characteristic cytoplasmic localisation of the different cytokines visualised with UV immunochemistry. Hence, the methods described here provide a reproducible, semiquantitative and specific assay for the detection of cell associated monokines. The technique may be applicable to the analysis of a variety of different cytokines in other phenotypically defined cell populations.