The identification of the determinants of bacterial interference with non-specific host defences during the early stages of infection is approached rather than attained. Recognizing a relevant biological property (e.g. resistance to phagocytosis) by an
in vitro
test and associating bacterial surface components with it are relatively easy. Proving causation, however, is usually not completed because the biological test is complex and the surface component(s) act only
in situ
. Nevertheless, evidence in addition to mere association can be sought to show that a putative determinant is strongly implicated in biological activity. Even then, proving that the biological activity concerned is relevant to infection
in vivo
, and that the putative determinant is produced there, is often not accomplished. Again, however, distinction can be made between those cases probably relevant
in vivo
and those only possibly so. Finally, bacteria grown
in vitro
can be deficient in some of the determinants of pathogenicity expressed during infection, and this situation requires the study of organisms grown
in vivo
. These points are discussed and then illustrated in a brief survey of the activities of many pathogenic bacteria and a description of recent work on the resistance of gonococci to killing by human serum and phagocytes.