Carbonaceous aerosols are present in many workplace and environmental settings. Some of these aerosols are known or suspect human carcinogens and have been linked to other adverse health effects. Exposure to diesel exhaust is of particular concern because it has been classified as a probable human carcinogen and use of diesel-powered equipment is widespread in industry. Because previously used methods for monitoring exposures to particulate diesel exhaust lack adequate sensitivity and selectivity, a new method was needed. A carbon analysis technique called the 'thermal-optical method' was evaluated for this purpose. In thermal-optical analysis, carbon in a filter sample is speciated as organic and elemental (OC and EC, respectively), When the thermal-optical method was initially evaluated, only one instrument was available, so interlaboratory variability could not be examined, More recently, additional instruments were constructed and an interlaboratory comparison was completed. Eleven laboratories participated in the study, including four in Europe that employ an alternative thermal technique based on coulometric detection of CO2, Good agreement (RSDs less than or equal to 15%) between the total carbon results reported by all laboratories was obtained, Reasonable within-method agreement was seen for EC results, but the EC content found by the two methods differed significantly. Disagreement between the OC-EC results found by the two methods was expected because organic and elemental carbon are operationally defined. With all filter samples, results obtained with the coulometric method were positively biased relative to thermal-optical results. In addition, the alternative method gave a positive bias in the analysis of two OC standard solutions. About 52% and 70% of the carbon found in two aqueous solutions containing OC only (sucrose and EDTA, respectively) was quantified as elemental, while EC contents of about 1% and 0.1% (respectively) were found by the thermal-optical method. The positive bias in the analysis of the OC standards is attributed largely to inadequate removal of OC during the first part of the analysis; lack of correction for pyrolytically formed carbon (char) also is a factor. Results obtained with a different thermal program having a higher maximum temperature were in better agreement with the thermal-optical method.