Statistical parameters with carbon felt dosimetry are analyzed and new sampling criteria are presented based on the analysis. In cass where the mean of measured time-weighted averages is kept two times the unbiased standard deviation below the maximum permissible concentration-time weighted average (MPC-TWA) and the values above MPC-TWA are observed with a frequency of less than 5%, compliance is determined with a confidence limit of 95% provided the biological half time of the contaminant is longer than 4 hours.