Field entry, published 2026-11-03T00:00:00.000Z, updated 2026-11-03T00:00:00.000Z. Group: The four levels of ISCO-08

Why does ISCO-08 use 1, 2, 3 and 4 digit codes?

ISCO-08 uses codes of one, two, three, and four digits because different questions need different levels of detail, and a single level could not serve them all. A one digit code answers a broad question, a two digit code answers a field level question, a three digit code answers a cluster level question, and a four digit code answers the most detailed question the classification allows. The varying lengths are not decoration. They are the mechanism that lets one classification produce broad totals and detailed breakdowns from the same underlying structure, without forcing every user to work at the finest level.

The problem a single length would create

Suppose the classification offered only four digit codes. Every table would then have 436 rows, every comparison would require handling hundreds of small figures, and broad patterns would be buried under detail. Suppose, instead, it offered only one digit codes. Every table would have 10 rows, broad patterns would be clear, but all finer differences would vanish, and questions about particular kinds of work could not be answered at all. Neither extreme works for a classification that must serve national statistics, international comparison, research, and policy discussion at the same time. Multiple lengths resolve the tension by letting the user choose the level that fits the question, while keeping all levels linked through the same nesting.

What each length is for

A one digit code is for the overall shape of employment: the balance between broad families of work. A two digit code is for comparing fields inside those families, such as comparing health professional work, as in 22 Health Professionals, with other professional families inside 2 Professionals. A three digit code is for comparing clusters of related occupations when a field level view is too coarse but a full detailed list is too long. A four digit code is for the finest comparison the system supports, where neighbouring occupations must be kept apart. None of these uses is more correct than the others in the abstract. Correctness depends on fit between the code length and the question being asked.

Why length and detail move together

Length and detail move together because each added digit marks a further division. The first digit divides all work into 10 families. The second digit divides those families into 43 parts. The third digit divides those parts into 130 parts. The fourth digit divides those parts into 436 parts. Longer codes therefore always describe smaller groups, and shorter codes always describe larger groups, provided the codes are read as prefixes of each other. That regularity is what makes aggregation possible: detailed figures can be added up to broader totals, because the longer codes that belong to a shorter code are known in advance. Without that regularity, adding detailed figures together would risk double counting or omission.

When shorter is better, and when it is not

Shorter codes are better when samples are small, when countries apply detailed definitions differently, or when the aim is a headline comparison that must remain readable. In those cases, detail adds noise rather than insight. Shorter codes are worse when the question turns on a distinction that only appears at a finer level, for example when two neighbouring detailed groups move in opposite directions and their combined total hides the change. Longer codes are better when the data can support them and the question requires them. Longer codes are worse when they create tiny, unstable figures that invite overinterpretation. The skill in using ISCO-08 lies in matching length to purpose, not in always preferring the longest code available.

A practical rule for choosing a length

Start with the question, not with the code. Write the question in one sentence, then ask what level of answer would actually settle it. If a broad family total settles it, use one digit. If a field comparison settles it, use two digits. If a cluster comparison settles it, use three digits. If only a detailed group settles it, use four digits. Then check that the data at that length are large enough and consistent enough to support the conclusion you want to draw. If they are not, step up one level rather than forcing a detailed answer from fragile figures. That rule keeps the four lengths working as a set of tools, each chosen deliberately, instead of as a ladder that must always be climbed to the top.

Sources checked

1. ILO ISCO-08

Back to the The four levels of ISCO-08 group or browse all occupations.