A short guide to assembling custom age bands from official United Nations data, and why the year and scenario labels matter.
Why a single population total is not enough
Two countries can have a similar number of residents and still have very different age structures. That difference shows up in classrooms, in the labour force, and in demand for health and long-term care. The usual way to measure it is to add complete age bands and then divide by the total population.
The United Nations World Population Prospects 2024 revision is the standard international source for this work. It covers 237 countries and areas. Figures through 2023 are historical estimates. Figures from 2024 onward, including mid-2026, are projections. The numbers cited below use the medium fertility variant and refer to 1 July 2026, unless noted otherwise.
The official picture in mid-2026
Under that medium variant, the world's population on 1 July 2026 is 8.301 billion, and the global median age is 31.1 years. The median age is the age that splits a population in half: one half is younger, one half is older. It is a compact summary, but it does not tell you whether the people outside the middle are mostly children or mostly older adults.
Country medians already show how wide that range is:
Niger has a median age of 15.7 years and a total fertility rate of 5.64 live births per woman.
Nigeria has a median age of 18.3 years, a projected population of 242.4 million, and a total fertility rate of 4.21.
India has a median age of 29.2 years and a projected population of 1.477 billion, still the largest national total.
The United States has a median age of 38.7 years and a projected population of 349.0 million.
China has a median age of 40.6 years, a projected population of 1.413 billion, and a growth rate of −0.22 percent.
Japan has a median age of 50.2 years, a projected population of 122.4 million, a total fertility rate of 1.23, and a growth rate of −0.56 percent.
Japan and Nigeria are useful extremes. One has a high median age and a fertility rate well below the conventional replacement level of about 2.1. The other has a low median age and fertility still above 4. A median age alone cannot tell you the share of children or the share aged 65 and older. For that, you have to add the relevant age groups.
How to assemble a custom age range
The UN publishes populations in standard five-year bands: 0–4, 5–9, 10–14, and so on, through 100+. A custom range is the sum of every complete band you want to include. Three ranges come up constantly:
Ages 0–14. Add 0–4, 5–9 and 10–14. This is the usual child share in international comparisons.
Ages 15–64. Add every band from 15–19 through 60–64. This is the conventional working-age group in UN tables, not a claim that everyone in that range is employed or that no one outside it works.
Ages 65 and older. Add every band from 65–69 through 100+. The 100+ group has no upper bound.
Once you have the count, divide it by the total population of the same country and year. The share lets you compare a small country with a large one. The count tells you how many people the share actually represents. Both are needed. A modest percentage of a very large population can still be a large number of people.
The same method works for other questions, as long as you only add complete bands. Ages 5–19, often used as a rough school-age span, is 5–9 plus 10–14 plus 15–19. Ages 20–34 is 20–24 plus 25–29 plus 30–34. The method cannot isolate a range such as ages 6–17, because those ages sit inside five-year bands. If you need a single year of age, use the UN's single-year files rather than rounding a five-year sum.
What the shares are for, and what they are not
Demographers often combine the child and older shares into a dependency ratio: the number of people aged 0–14 and 65+ for every 100 people aged 15–64. The ratio is a starting point for thinking about the balance between people who are typically in education or retirement and people in the conventional working ages. It is not a measure of who is economically dependent. Labour-force participation, health, and retirement ages vary, and age 65 is a comparable convention, not a universal threshold.
The same caution applies to school-age sums. Adding ages 5–19 is not the same thing as an enrolment count. Actual school starting ages, leaving ages, and attendance rates come from education statistics, not from the age distribution alone.
Keep the year and the scenario on the figure
A 2026 figure from World Population Prospects 2024 is a medium-variant projection, not a census count and not a forecast that is guaranteed to occur. The UN also publishes other variants with different fertility assumptions. If you later compare a 2023 estimate with a 2026 or 2050 projection, label both the year and the variant. Mixing an estimate with an unlabelled projection, or mixing two revisions of the same year, is a common source of apparent disagreement between sources.
Life expectancy and fertility help explain why two age structures look different, but they are not substitutes for the age counts. Japan's life expectancy at birth in this series is 85.1 years in 2026; Nigeria's is 55.0 years. Those gaps matter for the top of the distribution. Fertility and the size of past birth cohorts matter for the base and the middle. The direct way to measure any one of those parts is still to add the bands and divide by the total.
Sources
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
Age Structure Desk · last reviewed November 2026.