The problems encountered during a national survey of occupational hazards are discussed which almost exclusively deal with collection of information of chemical products. Hazardous substances were difficult to categorize in most cases due to inadequate labelling of trade name products. The problem is compounded when a given product is reformulated and repackaged several times by several different manufacturers, some of them unfamiliar with the nature of the raw materials that they receive and modify; about 30 to 40% of the items surveyed had been modified and repackaged and this created such problems that a chemical thesaurus was developed. In using this hierarchically arranged scheme, one can trace broader terms through successively more specific levels to narrower terms for chemical substances. Ignorance about chemical content in some cases was due to insufficient scientific bases for classification of ingredients and created arrangements whereby some specific identifying characteristics (e.g. boiling point of petroleum fractions) could be applied to ingredients. Manufacturers were reluctant to send "trade secret" information to contractors operating the survey but did send data directly to the government, which maintained a tight security system to protect privileged information. Data from 4636 factories in 67 different metropolitan areas yielded 50,000 formulations and an additional 30,000 which need to be further investigated. About 2700 of the 50,000 products contain Class I carcinogens and 3300 contain Class II. In most cases, the pigments implicated as strong carcinogens are labelled as trade secrets when they are incorporated into products. A central repository of ingredient information and trade names file is recommended. Some system of color coding of containers to indicate different levels of toxicity which might allow workers to decide whether they want to be exposed to such materials is suggested.