Because transmission of HIV to health care workers after needlestick injury has occurred mainly a result of deep insertion of large gauge needles, blood and viable mononuclear cells transferred after needlestick injury were measured. Needles of 20 to 27 gauge were filled with HIV-1 seropositive blood and inserted through extracorporeal human skin or parafilm covering physiologic saline solution modified Drabkin's solution, or culture medium, or inserted directly into one of these fluids, to a depth of one third of the needle length (0.5 inch) for 1 second. Volume of blood transferred was measured by both modified Drabkin's method and by chromium 51 labeling of red blood cells. Transfer of viable mononuclear cells was measured by growth in culture medium containing autologous feeder cells. The volume of blood transferred from a needle passed through skin varied from 312 +/- 69 nl from a 20-gauge needle to 14 +/- 4 nl from a 27-gauge needle, as measured by modified Drabkin's technique, and from 404 +/- 80 nl to 12 +/- 3.1 nl, as measured by chromium 51 labeling of red blood cells. The volume of blood transferred from a needle passed through parafilm was twice that transferred through skin. The volume of blood transferred through skin was 40% that transferred directly into fluid not covered by any barrier; blood transferred through parafilm was 80% of that transferred directly. When needles containing blood were inserted into culture medium for 1 second in the absence of a barrier, at least one viable mononuclear cell was almost always transferred to fluid from all gauges of needle tested. Insertion of needles through skin prevented transfer of all viable mononuclear cells from only 3% to 5% of 20- to 23-gauge needles, and from 12% to 32% of 26- and 27-gauge needles. Parafilm was an even less effective barrier than skin. Insertion of needles through parafilm completely prevented transfer of viable mononuclear cells from no 20- to 23-gauge needles and from only 5% to 10% of 26- and 27-gauge needles. The volume of blood transferred after needle insertion through skin for 1 second varied with the gauge of the needle and was 30-fold higher from a 20-gauge than from a 27-gauge needle. Variable mononuclear cells were transmitted after insertion through skin from more than 95% of 20- to 23-gauge needles and from 68% to 88% of 26- and 27-gauge needles. Parafilm was less effective than skin in reducing transmission of blood and viable mononuclear cells.