An experimental study has been conducted on the penetration of silicone rubbers and human skin in vivo by sharp-tipped and flat-bottomed cylindrical punches. A penetrometer was developed to measure the penetration of human skin in vivo, while a conventional screw-driven testing machine was used to penetrate the silicone rubbers. The experiments reveal that the penetration mechanism of a soft solid depends upon the punch tip geometry: a sharp tipped punch penetrates by the formation and wedging open of a mode I planar crack, while a flat-bottomed punch penetrates by the growth of a mode II ring crack. The planar crack advances with the punch, and friction along the flanks of the punch leads to a rising load versus displacement response. In contrast, the flat-bottomed punch penetrates by jerky crack advance and the load on the punch is unsteady. The average penetration pressure on the shank cross section of a flat-bottomed punch exceeds that for a sharp-tipped punch of the same diameter. In addition, the penetration pressure decreases as the diameter of the sharp-tipped punch increases. These findings are in broad agreement with the predictions of Shergold and Fleck [Proc. R. Soc. London, Ser. A (in press)] who proposed models for the penetration of a soft solid by a sharp-tipped and flat-bottomed punch.