Tissue sites involved in certain types of inflammation become sensitive to the destructive effect of a subsequent injection of tumour necrosis factor alpha (TNF alpha). To try to further delineate the cascade of effector and regulatory events controlling this activity, a new model is described and its main properties characterized. C57BL/GrFa mice received mycobacterial products subcutaneously in the footpads. Recombinant TNF alpha was injected 24 h later into the same sites. To assess the tissue-destructive effect of TNF alpha in these "primed" footpads, swelling and haemoglobin content of injected footpads were measured, 16 h and 20 h respectively after the injection of TNF alpha. When loaded with either Escherichia coli LPS (10 micrograms) or Mycobacterium vaccae soluble sonicate (17 micrograms), footpads were reactive to the subsequent injection of 1 microgram recombinant TNF alpha, as assessed by both swelling and haemoglobin content. When C57BL/GrFa mice received 10(9) autoclaved M. vaccae subcutaneously in the back 10 days before the footpad was "primed" with soluble M. vaccae sonicate, the destructive effect of TNF alpha was significantly enhanced, becoming 5-10-fold greater than that seen in sites "primed" with an optimal dose of LPS. This higher reactivity was abrogated by a single dose of anti-CD4 given just before the injection of TNF alpha. This local reactivity to TNF alpha of skin sites loaded with mycobacterial products is compared to the local LPS-dependent Shwartzman reaction, and the relevance of this assay as a model with which to delineate the mechanisms of tissue damage in tuberculosis is discussed.