How many categories are there at each level in Rev. 5?

ISIC Rev. 5, UN Statistics Division, contains 22 sections labelled A to V, 87 divisions, 258 groups and 463 classes, and those four totals are the headline counts for the revision. The totals describe the international structure only; national classifications derived from ISIC may add further detail below class level. Quoting the totals without the version label would invite confusion, because earlier revisions had different counts at each level.

Why do the four totals need to be quoted together?

The four Rev. 5 totals, 22 sections, 87 divisions, 258 groups and 463 classes, belong together because each total on its own invites a wrong inference. Quoting 463 classes alone might suggest a flat list of 463 items. Quoting 22 sections alone might suggest a classification with only broad brush strokes. Together, the four numbers describe a shape: broad at the top, progressively finer below, with far more detail at class level than at section level. A writer who quotes one total should therefore quote the set, with the version label, ISIC Rev. 5, UN Statistics Division, attached, so a later reader knows which edition the counts describe. Earlier revisions had different counts at each level, and a total carried across revisions without its label is a total that can no longer be checked.

That shape is also uneven. Not every section splits into the same number of divisions, and not every division splits into the same number of groups and classes. Some fields of activity are divided finely because statistical users need the detail, while other fields stay broader because the work inside them is more uniform or less in need of finer breakdown for statistical purposes. The totals therefore describe the international structure as a whole, not a promise that every branch of the tree is the same depth or the same width. A reader who expects symmetry will be puzzled by branches that stop early. A reader who expects a working tool will see branches shaped by need.

How do national versions relate to these international totals?

National classifications derived from ISIC may add further detail below class level, and that extra detail does not change the international totals. A national version can split a class for domestic purposes, or add categories beneath it, while still mapping back to the international structure above. The relationship runs in one useful direction: national detail refines the international class, but the international totals remain 22, 87, 258 and 463 regardless of how much extra detail any one country adds. A writer who borrows a count from a national publication should therefore ask whether the count describes the international structure or the national extension, because the two counts answer different questions.

That distinction also protects comparisons between countries. Two countries may publish at different levels of national detail, yet both can be compared at the international levels they share, provided the version label and the level are stated. A table that mixes a national subdivision from one country with an international class from another, without saying so, creates an apparent difference that is really a difference in labelling. Naming the structure, the version and the level in the methods note prevents that false difference from entering the argument.

What should a writer do when a total from another revision appears beside Rev. 5?

First, do not silently substitute one label for another. Where Rev. 5 is meant, name ISIC Rev. 5, UN Statistics Division, and where an earlier revision is meant, name that revision instead. Second, check the reference period, because revisions and national adoption dates can fall inside a series, and a series that crosses a revision boundary needs its break flagged. Third, check the level, because a change in counts at one level does not imply the same change at every level: categories may have been split, merged or moved at class level while the section total stayed the same, or the reverse. Each of those moves has a different effect on a comparison, and only the UNSD structure for the relevant revision shows which move actually happened.

A short checklist, kept with the working notes, covers most cases: version label written down, level stated, reference period noted, source named, and any break in series flagged. The list takes minutes. Correcting a published comparison that omitted it takes far longer, and the correction often reaches fewer readers than the original error.

This guide counts categories in a statistical classification only, and it assigns no reader to any category. It does not determine a reader’s code, principal activity, or registration, and a code for a specific record needs confirmation from the national statistical institute or business register that holds the record.