Sampling bias in full facepiece respirators was evaluated. A MSA, two Scott brand, and a US Divers full face respirator were tested. The effects of face seal leakage (at the corner of the eye, cheek, and chin), probe location (on the midline opposite the mouth, in the area between the nose and mouth, and near the nose), and breathing pattern on sampling bias were evaluated. The respirators were mounted on a mannequin and acetone was used as the test agent. Probe location and depth, area of face seal leakage, and nose or mouth breathing were the major factors contributing to sampling bias. For example, in the MSA respirator the mean sampling bias with the probe being located near the mouth was -37 percent. This increased to -46 and -75 percent when the probe was in the mid/nose/mouth and nose positions, respectively. Significantly better sampling occurred in one of the Scott respirators when the face seal leakage occurred near the eyes relative to the other positions, regardless of probe location. Relatively small degrees of sampling bias were found with the US Divers respirator. The least bias was found with the probe located near the mid/nose/mouth position. Nose breathing produced less bias than mouth breathing. The authors conclude that in/facepiece sampling from full facepiece respirators leads to unrepresentative sampling resulting from location of the sampling probe, area where the face seal leak occurs, breathing pattern, and facepiece design. The in/facepiece sampling procedure is not suitable for providing reliable quantitative performance data on full facepiece respirators.