EPSO publishes a handful of sample numerical questions, and they have one thing in common: none of them can be answered by reading a single cell. The GDP ratio has to be built from R&D spending and R&D intensity. The town’s population has to be worked out from its total waste and its waste per head. The eight questions below are built the same way, on real Eurostat data, with the answer format EPSO uses.
What EPSO’s numerical questions look like
Each question comes with its own table or chart, usually with more rows and columns than you need. The stem asks for one number, ratio or country, and there are four or five options. The last option is sometimes “None of the above”, and it is sometimes right. You can see EPSO’s own samples on its numerical reasoning page.
In the current AD5 competition (notice EPSO/AD/427/26) the numerical test has 10 questions in 20 minutes, and it shares its pass mark with the abstract reasoning test: you need 10 out of 20 across the two tests combined. A strong numerical score can carry a weak abstract one, and the other way round. For the format and the recurring traps in general, see our numerical reasoning guide; this page is for practice.
The eight questions
Every number in the tables is Eurostat data, rounded as shown. Work from the table exactly as printed: the answers are calculated from the displayed values. The questions go from medium to very hard.
GDP per head and population, selected Member States
| Country | GDP per head (€) | Population, 1 Jan (millions) | ||
|---|---|---|---|---|
| 2024 | 2025 | 2024 | 2025 | |
| Germany | 52,530 | 54,240 | 83.46 | 83.58 |
| Ireland | 104,350 | 110,150 | 5.35 | 5.44 |
| Netherlands | 61,980 | 64,710 | 17.94 | 18.04 |
| Poland | 22,710 | 24,700 | 36.62 | 36.50 |
| Portugal | 27,220 | 28,560 | 11.20 | 11.39 |
| Sweden | 53,010 | 56,250 | 10.55 | 10.59 |
Data: Eurostat, nama_10_pc, tps00001 (retrieved 29 September 2026).
- 2:9
- 2:3
- 3:2
- 7:1
- None of the above
Show answer
• Poland: 24,700 × 36.50 million = €901.6 billion
• Ireland: 110,150 × 5.44 million = €599.2 billion
• 901.6 ÷ 599.2 = 1.50, which is 3:2.
Why the other options are wrong.
• A, 2:9 is the ratio of GDP per head alone (24,700 ÷ 110,150 = 0.224). It forgets that Poland has almost seven times as many people.
• B, 2:3 is the right calculation divided the wrong way round: Ireland to Poland.
• D, 7:1 compares population only (36.50 ÷ 5.44 = 6.71).
• E is for candidates who expect Ireland, with the highest GDP per head in the table, to have the larger economy and cannot find a matching option.
20-second shortcut. Split the ratio into its two parts: population about 6.7 to 1, GDP per head about 1 to 4.5. 6.7 ÷ 4.5 ≈ 1.5. The 2024 columns and the four other countries are there to be ignored.
Municipal waste, selected Member States
| Country | Waste per inhabitant, 2013 (kg) | Waste generated, 2023 (thousand tonnes) | Population on 1 January 2023 (millions) |
|---|---|---|---|
| Denmark | 813 | 4,513 | 5.93 |
| Netherlands | 526 | 8,316 | 17.81 |
| Austria | 578 | 7,142 | 9.10 |
| Poland | 297 | 13,448 | 36.75 |
| Portugal | 439 | 5,338 | 10.93 |
Data: Eurostat, env_wasmun, tps00001 (retrieved 29 September 2026).
- 36 kg
- 207 kg
- 785 kg
- 6,564 kg
- None of the above
Show answer
• 7,142 thousand tonnes = 7,142,000,000 kg
• 7,142,000,000 kg ÷ 9,100,000 people = 784.8 kg per inhabitant in 2023
• 784.8 − 578 = 206.8 → 207 kg
Why the other options are wrong.
• A, 36 kg is the percentage increase (206.8 ÷ 578 = 36%) written with the wrong unit.
• C, 785 kg is the 2023 level. It answers “how much”, not “by how much did it increase”.
• D, 6,564 kg subtracts 578 kg per person from 7,142 thousand tonnes: two different units, so the result means nothing.
• The other four countries are there to be ignored: the question is only about Austria.
20-second shortcut. Thousand tonnes divided by millions of people gives kilograms per person directly, because a thousand tonnes is a million kilograms. 7,142 ÷ 9.1 ≈ 785, minus 578 ≈ 207. Learn this unit pair: it appears in EPSO’s own sample question on waste.
Arrivals and nights spent at tourist accommodation (millions)
| Country | Arrivals | Nights spent | ||
|---|---|---|---|---|
| 2023 | 2024 | 2023 | 2024 | |
| Greece | 35.8 | 37.2 | 147.2 | 152.9 |
| Spain | 144.8 | 149.7 | 485.0 | 505.2 |
| France | 180.7 | 181.1 | 460.3 | 457.6 |
| Croatia | 19.5 | 20.2 | 92.3 | 93.6 |
| Italy | 133.6 | 139.6 | 447.2 | 466.2 |
| Netherlands | 49.4 | 51.3 | 142.3 | 145.4 |
| Austria | 40.7 | 42.2 | 127.8 | 131.2 |
| Portugal | 30.8 | 32.3 | 84.9 | 88.1 |
Data: Eurostat, tour_occ_arnat, tour_occ_ninat (retrieved 29 September 2026). Hotels, holiday and other short-stay accommodation, campsites; all guests.
- Greece
- Spain
- Croatia
- Italy
- Portugal
Show answer
Pass 1, growth in nights above 3%: Greece +3.87%, Spain +4.16%, France −0.59%, Croatia +1.41%, Italy +4.25%, Netherlands +2.18%, Austria +2.66%, Portugal +3.77%. Only Greece, Spain, Italy and Portugal pass.
Pass 2, nights ÷ arrivals in 2024: Greece 152.9 ÷ 37.2 = 4.11; Spain 505.2 ÷ 149.7 = 3.37; Italy 466.2 ÷ 139.6 = 3.34; Portugal 88.1 ÷ 32.3 = 2.73. Greece, 4.11 nights, is the longest.
Why the other options are wrong.
• Croatia has the longest stay in the whole table (4.63 nights), but its nights grew by only 1.41%. It is the answer for anyone who skips the first condition, which makes it the most tempting wrong option.
• Spain has the most nights. A big total says nothing about the length of a stay.
• Italy grew fastest (+4.25%, just ahead of Spain’s +4.16%), which answers the first condition, not the question.
• Portugal is what you get by dividing arrivals by nights: the largest value of that inverted ratio is the shortest stay.
20-second shortcut. For the filter, 3% of 92.3 is about 2.8, so Croatia needed about 95.1 and reached only 93.6: out, without a calculator. For the ranking, only Greece has nights above four times its arrivals (4 × 37.2 = 148.8 < 152.9).
Share of renewable energy in gross final energy consumption (%)
Data: Eurostat, nrg_ind_ren (retrieved 29 September 2026).
- 63.7%
- 70.3%
- 73.8%
- 75.6%
- None of the above
Show answer
Step 1, percentage points (the difference): Belgium 6.3; Denmark 17.2; Ireland 7.6; Netherlands 14.8; Portugal 6.8; Finland 13.5. Denmark rose the most, by 17.2 points.
Step 2, percentage (the ratio): Denmark went from 29.3 to 46.5, a factor of 46.5 ÷ 29.3 = 1.5870, or +58.7%.
Step 3: 46.5 × 1.5870 = 73.80 → 73.8%.
Why the other options are wrong.
• A, 63.7% adds another 17.2 points. That repeats the same percentage-point rise, not the same percentage.
• B, 70.3% projects Finland, the country with the highest share, which is not what the question asks for.
• D, 75.6% projects the Netherlands, whose share almost quadrupled. It had the largest rise in per cent (+274%) but only 14.8 points. This is the classic points-versus-per-cent trap, and the chart invites it: the Dutch bar quadruples, the most striking change on the page.
20-second shortcut. Read the points off the labels (Denmark 17.2, Netherlands 14.8, Finland 13.5). For the projection, the factor 46.5 ÷ 29.3 is just under 1.6, so the answer is just under 46.5 × 1.6 = 74.4. That is C, not D. Growth “by the same percentage” always means multiplying by the same factor again.
Monthly minimum wage in national currency, and euro exchange rates
| Country | Currency | Monthly minimum wage | Average exchange rate 2025 (units per €1) | |
|---|---|---|---|---|
| July–Dec 2025 | Jan–June 2026 | |||
| Czechia | CZK | 20,800 | 22,400 | 24.688 |
| Hungary | HUF | 290,800 | 322,800 | 397.77 |
| Poland | PLN | 4,666 | 4,806 | 4.2397 |
| Romania | RON | 4,050 | 4,050 | 5.0424 |
Data: Eurostat, earn_mw_cur, ert_bil_eur_a (retrieved 29 September 2026).
- 1,134 CZK
- 22,400 CZK
- 27,170 CZK
- 27,986 CZK
- 118,651 CZK
Show answer
• Złoty to euro: 4,806 ÷ 4.2397 = €1,133.57
• Euro to koruna: 1,133.57 × 24.688 = 27,985.6 → 27,986 CZK
Rates are quoted as units per €1, so you divide to get euros and multiply to leave them.
Why the other options are wrong.
• A, 1,134 stops halfway: that is the wage in euros, not koruna.
• B, 22,400 is Czechia’s own minimum wage for the same period, read straight from the table.
• C, 27,170 converts the July–December 2025 wage (4,666 PLN). Check the column before you check the arithmetic.
• E, 118,651 multiplies złoty by the koruna rate and skips the first step.
20-second shortcut. The cross rate is 24.688 ÷ 4.2397 ≈ 5.8 koruna per złoty, and 4,806 × 5.8 ≈ 27,900. That rules out everything but C and D, and the column rules out C.
Population on 1 January 2025, and change
| Country | Population (millions) | Change since 1 January 2024 (%) | Change since 1 January 2015 (%) |
|---|---|---|---|
| Ireland | 5.44 | +1.7 | +16.3 |
| Greece | 10.37 | 0.0 | −4.1 |
| Italy | 58.94 | 0.0 | −2.2 |
| Poland | 36.50 | −0.3 | −4.0 |
| Portugal | 11.39 | +1.6 | +9.5 |
| Romania | 19.04 | −0.1 | −4.2 |
| Sweden | 10.59 | +0.3 | +8.6 |
Data: Eurostat, tps00001 (retrieved 29 September 2026).
- 14.9
- 15.1
- 16.6
- None of the above
Show answer
• Ireland: 5.44 ÷ 1.163 = 4.678
• Portugal: 11.39 ÷ 1.095 = 10.402
• Together: 15.079 → 15.1 million
Why the other options are wrong.
• A, 14.9 takes 16.3% and 9.5% off the 2025 figures (5.44 × 0.837 + 11.39 × 0.905). It is the most common mistake in this question type. Averaging the two percentages and applying 12.9% to the total lands on the same wrong number.
• C, 16.6 uses the wrong column: the change since 2024, not 2015.
• D is for candidates who calculate correctly but round too early. Keep three decimals until the last step.
20-second shortcut. Dividing by 1.163 takes off a little less than 16.3%, so the right answer is always above the “subtract the percentage” answer. If both appear among the options, the higher one is the reverse-percentage answer.
Unemployment rate, people aged 15–74 (% of the labour force)
Data: Eurostat, une_rt_a (retrieved 29 September 2026).
- 7%
- 46%
- 50%
- 117%
- None of the above
Show answer
Why the other options are wrong.
• A, 7% is the fall in percentage points, reported as a percentage.
• B, 46% is how much of the 2015 rate is left (6.0 ÷ 13.0), not how much it fell.
• C, 50% is the eyeball answer: 13 to 6 “looks like” halving. The question asks for the nearest whole per cent, so an estimate is not enough.
• D, 117% divides by the new value instead of the old one.
20-second shortcut. Half of 13.0 is 6.5; the rate ended below half, so the fall is more than 50%. That kills B and C at once, and A and D are unit and base errors. Choose “None of the above” with confidence when your number is clean and every option has a name.
Gross domestic product and research and development (R&D) spending
| Country | GDP (€ billion) | R&D spending (% of GDP) | ||
|---|---|---|---|---|
| 2019 | 2024 | 2019 | 2024 | |
| Czechia | 229.4 | 321.2 | 1.90 | 1.82 |
| Greece | 185.2 | 236.7 | 1.26 | 1.54 |
| Spain | 1,253.7 | 1,594.3 | 1.24 | 1.50 |
| Hungary | 147.4 | 206.2 | 1.46 | 1.31 |
| Poland | 538.4 | 852.2 | 1.31 | 1.41 |
| Portugal | 214.5 | 291.2 | 1.39 | 1.73 |
| Romania | 223.3 | 353.6 | 0.48 | 0.46 |
Data: Eurostat, nama_10_gdp, rd_e_gerdtot (retrieved 29 September 2026). GDP at current prices.
- 8%
- 58%
- 66%
- 70%
- None of the above
Show answer
• 2019: 538.4 × 1.31% = €7.05 billion
• 2024: 852.2 × 1.41% = €12.02 billion
• 12.02 ÷ 7.05 = 1.7037 → +70%
Why the other options are wrong.
• A, 8% is the growth of the R&D share alone (1.41 ÷ 1.31). It ignores that the economy it is a share of also grew.
• B, 58% is GDP growth alone.
• C, 66% adds the two growth rates (58.3 + 7.6). Growth rates multiply: 1.583 × 1.076 = 1.704. The gap between adding and multiplying is the product of the two rates (0.583 × 0.076 ≈ 0.044), which is exactly what separates C from D.
20-second shortcut. Whenever a quantity is “A × B” and both change, multiply the factors and never add the percentages. Here 1.58 × 1.08 ≈ 1.71: D.
These questions are written by EU Testing in the EPSO format. They are not official EPSO questions.
Eight questions, eight traps
Every wrong option above comes from a real mistake, and the same few mistakes come back in every EPSO numerical test. The skill is to recognise the question type in the first ten seconds:
| Question | What it tests | The trap to expect |
|---|---|---|
| 1 | A ratio of values you must build first | Comparing the per-head figures alone |
| 2 | A change per head, from a total | Mixing units: thousand tonnes and kilograms |
| 3 | A condition, then a ranking | Skipping the condition |
| 4 | Percentage points versus per cent | Choosing the biggest bar instead of the biggest difference |
| 5 | Converting between two non-euro currencies | Stopping at euros, or using the wrong period |
| 6 | Reverse percentage | Subtracting the percentage from today’s value |
| 7 | “None of the above” | Accepting a rounded eyeball answer |
| 8 | Two changes that compound | Adding growth rates instead of multiplying |
Two habits cover most of them. First, before you calculate anything, say what the answer is made of: “GDP equals GDP per head times population”, “R&D in euros equals GDP times the R&D share”. Second, estimate before you press a key. In almost every question above, a rough estimate rules out three options, and the calculator only has to separate the last two.
60 more questions in this style
The Numerical Reasoning Focus card has 60 questions on real EU data, from medium to very hard, each with a worked explanation that names the trap.
Open the Numerical Focus card → First 5 questions free · all 60 with PremiumFrequently asked questions
How many questions are in the EPSO numerical reasoning test?
In the AD5 competition (EPSO/AD/427/26), 10 questions in 20 minutes, scored from 0 to 10. There is no separate pass mark: numerical and abstract reasoning share a combined pass mark of 10 out of 20.
Is “None of the above” ever the right answer?
Yes. EPSO’s own sample includes it as an option, and a well-built test makes it correct often enough that you cannot ignore it. Choose it when your calculation is clean and every other option matches a mistake you can name, as in question 7 above.
Can I use a calculator?
Yes, EPSO provides an on-screen calculator. It is slower than a physical one, so estimating first and calculating only to separate the last two options saves time.
Do I need to learn formulas?
Only a few: percentage change (new minus old, divided by old), reverse percentage (divide by one plus the change), per head (total divided by population) and currency conversion (divide by the rate to get euros, multiply to leave them). The questions test whether you apply them to the right numbers.
Where do the data in these questions come from?
From Eurostat, the EU’s statistical office, retrieved on 29 September 2026. Each table names the Eurostat dataset it uses. The figures are rounded for display, and the answers are calculated from the rounded values shown.
Data: Eurostat (datasets named under each table), retrieved 29 September 2026. Pass marks: EPSO notice of competition EPSO/AD/427/26.