GHS compilation is the process of gathering toxicological data, physical hazard information, and environmental fate data for a chemical substance or mixture, then organizing it into the standardized format required by the Globally Harmonized System of Classification and Labelling of Chemicals. The system itself grew out of a 1992 recommendation at the Rio Earth Summit, which recognized that the multiplicity of existing national and regional hazard classification schemes was creating dangerous confusion for workers, consumers, and emergency responders worldwide.1Journal of Hazardous Materials. The development of the globally harmonized system (GHS) of classification and labelling of hazardous chemicals For anyone responsible for chemical safety in a workplace, understanding GHS compilation means understanding what data you need, how to classify it correctly, and how the final output (labels, safety data sheets, and pictograms) must look.
What GHS Compilation Actually Covers
At its core, GHS compilation is the act of taking raw hazard data about a chemical and translating it into the system’s standardized language. That language consists of four key elements: signal words (either “Danger” or “Warning”), pictograms (the familiar red-bordered diamond symbols), hazard statements (short phrases describing the nature of the hazard), and precautionary statements (advice on handling, storage, and emergency response).2Elsevier (Procedia Engineering). Understanding a Safety Data Sheet (SDS) in Regards to Process Safety A compiled GHS entry for a single substance might draw from dozens of toxicological studies, physical property tests, and environmental data points to arrive at the correct combination of these elements.
The hazards themselves fall into three broad domains. Physical hazards cover things like flammability, explosivity, oxidizing properties, and reactivity with water. Health hazards address what happens when a person is exposed. Environmental hazards deal with a chemical’s toxicity to aquatic life and its persistence in ecosystems. Each domain has its own set of categories, and the compiler’s job is to match the available data to the correct category for each relevant endpoint.
Health Hazard Endpoints
Health hazards tend to be the most data-intensive part of GHS compilation. The system evaluates chemicals across several toxicological endpoints: acute toxicity (how dangerous a single exposure is), irritation or corrosivity (damage to skin or eyes), sensitization (whether a chemical triggers allergic responses), carcinogenicity, mutagenicity, reproductive toxicity, and chronic or repeat-dose toxicity.3Toxicology Letters. Global harmonisation of classification and labelling of hazardous chemicals Each endpoint has its own classification criteria and category structure, so a single chemical can carry multiple health hazard classifications simultaneously. A solvent, for instance, might be classified as acutely toxic by inhalation, a skin irritant, and a suspected carcinogen, with each classification generating its own pictogram, hazard statement, and precautionary advice.
The carcinogenicity classification illustrates how the category system works in practice. GHS distinguishes between Category 1, which covers chemicals known or presumed to cause cancer in humans, and Category 2, which covers suspected human carcinogens where the evidence is suggestive but not conclusive. Category 1 is further split: 1A applies when the evidence comes largely from human epidemiological studies, while 1B applies when the evidence rests primarily on animal data.4PubMed. Guidance for the classification of carcinogens under the Globally Harmonised System of Classification and Labelling of Chemicals (GHS) A compiler needs to weigh the available evidence against these criteria, and the distinction matters because it affects which signal word appears on the label and how downstream users are expected to handle the substance.
Classifying Mixtures Versus Pure Substances
Classifying a pure substance with solid test data is relatively straightforward. Mixtures are where GHS compilation gets genuinely difficult. Most commercial products are mixtures, from cleaning sprays to industrial coatings, and testing every mixture as a whole is often impractical or prohibitively expensive. The GHS addresses this through a tiered approach: if you have test data on the mixture itself, use it first; if not, use data on similar tested mixtures through bridging principles; and if neither is available, calculate the mixture’s hazard classification from the known hazards of its individual ingredients.
That calculation step uses what is commonly called the GHS mixtures equation, which estimates acute toxicity by combining the toxicities and concentrations of each ingredient. In practice, this formula works better for some product types than others. One large evaluation found that when using U.S. EPA classification categories, the equation’s predictions matched actual test results about 55% of the time across a dataset of 671 formulations. The accuracy varied sharply by product type: agrochemical formulations showed only 52% concordance, while antimicrobial cleaning products hit 84%.5PubMed Central. Performance of the GHS Mixtures Equation for Predicting Acute Oral Toxicity When the analysis was restricted to formulations with relatively low toxicity (above 500 mg/kg), concordance improved to about 82%. The takeaway for compilers is that the equation is a reasonable screening tool, but it can misclassify products, particularly highly toxic ones, and professional judgment still matters.
The Safety Data Sheet as the Final Product
For most people doing GHS compilation, the tangible end product is the Safety Data Sheet, the 16-section document that accompanies every hazardous chemical in commercial use. The SDS is organized in a fixed order mandated by the GHS: identification, hazard identification, composition, first-aid measures, firefighting measures, accidental release, handling and storage, exposure controls, physical and chemical properties, stability and reactivity, toxicological information, ecological information, disposal considerations, transport information, regulatory information, and other information. Each section must contain specific types of data, and the compiler’s responsibility is to ensure every section is accurate and complete.
Because a single chemical manufacturer might produce hundreds or thousands of substances and mixtures, each requiring its own SDS in multiple languages, compilation has increasingly moved toward software-assisted workflows. Modern systems can ingest SDS data from various formats, apply validation layers to check for completeness and GHS compliance, and flag inconsistencies between the classified hazards and the pictograms shown on the document.6PubMed Central. Mitigation of Chemical Reporting Liabilities through Systematic Modernization of Chemical Hazard and Safety Data Management Systems These tools use techniques like computer vision to verify that the pictogram images on a processed SDS actually match the hazard codes assigned to the product, catching errors that a human reviewer might miss in high-volume operations.
Why the Same Chemical Can Have Different Labels in Different Countries
One of the most confusing aspects of GHS compilation is that “globally harmonized” does not mean “globally identical.” The GHS is a voluntary framework published by the United Nations. Individual countries and regions adopt it into their own regulations, and they can choose which hazard categories to include and which to leave out. They can also implement different editions of the GHS at different times, since the UN publishes revised editions on a regular cycle.
The European Union took a mandatory approach, incorporating GHS elements into its CLP Regulation (EC No 1272/2008) in 2008, which covers all GHS hazard classes and categories for substances and mixtures. The World Health Organization, by contrast, used a voluntary approach, folding GHS elements into its recommended classification of pesticides by hazard in 2009. Analysis of these two approaches found that the mandatory EU framework covered all GHS elements referenced in the second revised edition of the system, while the voluntary WHO approach was more selective.7J-STAGE / Industrial Health. A Comparison of Mandatory and Voluntary Approaches to the Implementation of Globally Harmonized System of Classification and Labelling of Chemicals (GHS) in the Management of Hazardous Chemicals Other countries, such as the United States (through OSHA’s Hazard Communication Standard), Japan, South Korea, Australia, and China, each have their own implementing legislation with their own quirks.
For companies that ship chemicals internationally, this fragmented adoption means that a single product may need multiple GHS-compiled labels and SDS documents, each tailored to the regulatory requirements of the destination country. The compilation process has to track which GHS revision edition each country follows, which “building blocks” (optional hazard categories) that country has adopted, and any country-specific concentration cutoff values for mixture classification. It is harmonization with an asterisk.
Environmental Hazard Classification and Emerging Concerns
The environmental hazard portion of GHS compilation focuses primarily on aquatic toxicity, both acute and chronic, as well as bioaccumulation potential and environmental persistence. These classifications determine whether a product carries the “environmental hazard” pictogram (the dead tree and fish symbol) and which precautionary statements must appear on the label. For many industrial and consumer chemicals, the environmental classification is where the most contentious data questions arise.
A good example of why this matters is the ongoing scrutiny of water-repellent and waterproofing chemistries. Over the past fifteen years or so, concerns about bioaccumulation and long environmental persistence have put intense pressure on specific chemical families used in these products. Long-chain fluorochemicals like PFOA and PFOS, as well as certain cyclic siloxanes, have attracted the most regulatory attention. Reviews of available data suggest that hydrocarbon-based polymers are the most environmentally benign among the major classes of water-repellent ingredients, followed by siloxane-based and then short-chain PFAS-based polymers.8ScienceDirect (Woodhead Publishing). Waterproof and Water Repellent Textiles and Clothing For GHS compilers, such evolving science means that environmental classifications can shift as new persistence and bioaccumulation data emerge, sometimes reclassifying substances that were previously considered low-risk.
In Silico Models and the Future of Hazard Data
Traditionally, GHS classification for acute toxicity relied on animal test data, particularly rat oral LD50 studies. That approach is expensive, time-consuming, and increasingly challenged on ethical grounds. A significant shift is underway toward computational methods that can predict a chemical’s toxicity from its molecular structure alone.
One prominent tool is the Collaborative Acute Toxicity Modeling Suite, or CATMoS, which uses computational modeling to predict rat acute oral toxicity. It was developed with the explicit goal of reducing animal testing during the registration of new pesticide active ingredients.9PubMed Central. Evaluation of in silico model predictions for mammalian acute oral toxicity and regulatory application in pesticide hazard and risk assessment The appeal for GHS compilation is obvious: if a reliable computer model can predict where a new substance falls on the acute toxicity scale, you can classify it without conducting a new animal study.
Recent evaluations across multiple sectors suggest that current computational models provide highly predictive results, often generating classifications that are accurate or more conservative (meaning they err on the side of caution) across thousands of chemicals. Advocates have begun pushing for these models to be formally recognized within the GHS classification criteria themselves, arguing that combining computational predictions with expert review could eliminate the need for routine animal acute oral toxicity studies while still maintaining robust hazard identification.10PubMed. In silico acute oral toxicity models are fit for inclusion in GHS and DG Model regulations classification criteria For compilers, this shift means that the data inputs for GHS classification are evolving: instead of always starting with a stack of animal study reports, you may increasingly start with a computational prediction and supporting literature, particularly for new substances without extensive test histories.
The Pictogram Comprehension Problem
GHS compilation produces labels that are supposed to communicate hazards universally, across languages and literacy levels. The red diamond-bordered pictograms, featuring symbols like a skull and crossbones for acute toxicity or an exclamation mark for irritants, are designed to be immediately understood. The reality is messier.
Research on how people actually interpret GHS pictograms in real-world settings has raised serious questions about their effectiveness, particularly in developing countries. A study of South African farm workers found that the pictograms were frequently misunderstood, especially those representing more abstract concepts like chronic health hazards (the “health hazard” silhouette showing damage to internal organs). The researchers pointed out that the GHS pictograms were not pilot-tested with end users before the system was adopted, and that they encode complex risk assessment data, such as chronic versus acute toxicity, into simple images that many users cannot decode correctly.11PubMed. South African farm workers’ interpretation of risk assessment data expressed as pictograms on pesticide labels This is a challenge that no amount of careful compilation can fix on its own. A perfectly compiled GHS label is still only as effective as the user’s ability to read it.
The problem is not limited to low-literacy settings. Even in countries with high rates of formal education, studies have found that many workers cannot reliably distinguish between the nine GHS pictograms or correctly match them to the hazard they represent. The skull and crossbones is widely recognized, but the difference between the exclamation mark (less severe health hazards) and the health hazard silhouette (serious chronic hazards like carcinogenicity) trips up a large share of people. For anyone responsible for GHS compilation within an organization, this gap between label quality and label comprehension is a practical concern that should drive investment in training, not just document accuracy.
Common Mistakes in GHS Compilation
People compiling GHS documents for the first time tend to make a few predictable errors. One of the most common is applying classification criteria from the wrong edition of the GHS or from a different country’s implementing regulation. Since the EU’s CLP, OSHA’s HazCom, and Japan’s Industrial Safety and Health Act all implement different editions with different building-block choices, copying a classification from one jurisdiction and pasting it into an SDS for another can produce incorrect labels.
Another frequent issue is ingredient concentration cutoffs for mixtures. The GHS specifies threshold concentrations above which an ingredient’s hazard classification triggers a classification for the whole mixture, but these cutoffs differ by hazard class and sometimes by jurisdiction. A mixture containing 0.5% of a skin sensitizer might trigger a mixture classification in one country but not another, depending on which cutoff that country adopted. Compilers who do not track these variations produce SDS documents that are technically non-compliant even when the underlying toxicology data is correct.
A subtler trap involves the “bridging principles” used to classify untested mixtures based on similar tested ones. The GHS allows you to classify a new mixture by analogy if it is substantially similar to a mixture for which test data exists, but the criteria for “substantially similar” require careful judgment. Changing a fragrance component in an otherwise identical cleaning product might not change the acute toxicity classification, but replacing a surfactant with a different chemical class could change it dramatically. The rules provide a framework, but they leave room for interpretation that can lead to inconsistent results across companies.
Regulatory Convergence and Ongoing Harmonization Gaps
The broader story of the GHS is one of slow, uneven convergence. The EU’s REACH and CLP regulations together form one of the most comprehensive implementations, covering not just classification and labeling but also chemical registration and restriction. These frameworks have reshaped how chemical hazards are communicated to workers, consumers, and emergency responders throughout Europe.12Royal Society of Chemistry. Chemical Regulation at the European Level: Safeguarding Consumer Health and Protecting the Environment Other major economies have implemented their own versions, but the patchwork of different adopted editions and optional building blocks means that a truly global compilation, one set of documents valid everywhere, remains more aspiration than reality.
The UN publishes a revised edition of the GHS roughly every two years, and each revision can introduce new hazard classes, refine existing classification criteria, or adjust the guidance on mixture classification. Countries then decide whether and when to incorporate these changes into their own regulations. At any given moment, different jurisdictions may be operating under different revision editions, creating a moving target for multinational companies. The practical implication for compilers is that GHS compilation is not a one-time project. It requires ongoing monitoring of regulatory updates across every market a company sells into, along with periodic re-evaluation of existing classifications as new data or new edition criteria become available. The tools and automated systems that handle SDS authoring need regular updates as well, since a system configured for one revision edition will produce non-compliant output when a jurisdiction upgrades to a newer one.