Eurostat Daily Smokers of Cigarettes (sdg_03_30)

It is Eurostat's Sustainable Development Goal indicator on smoking prevalence, titled "Smoking prevalence by sex", and it sits inside the European Statistical System's SDG monitoring stack, where it backs SDG 3.4 tracking on premature mortality from non-communicable diseases. The definition is narrow on purpose: the indicator counts the share of the population aged 15 and over who report smoking boxed cigarettes, cigars, cigarillos or a pipe. Electronic cigarettes and snuff are explicitly excluded, so the rise of vaping cannot blur the combustible trend this series exists to measure.

The numbers come out of Eurobarometer surveys: self-reported answers collected face-to-face in respondents' homes across participating countries. Knowing the method tells you two things before you model anything. First, observations exist only for the years the survey actually ran, so the series is deliberately intermittent rather than continuously interpolated. Second, the question wording has been held stable across waves, which is what makes a 2006-versus-2023 comparison legitimate.

The result is one compact, authoritative table for the Tobacco industry: 35 European geopolitical entities, three sex breakdowns, percentages of population, roughly 620 observations spanning seven survey waves from 2006 to 2023. Get a sample of this dataset and rows come back cut to the countries, sexes and waves you name.

Rows exactly as they arrive:

# one observation = geography x sex x survey year, population aged 15+
dataset      : sdg_03_30            # Smoking prevalence by sex
unit         : PC_POP               # percentage of total population
age          : Y_GE15               # 15 years or over

geo          : EU27_2020            # EU-27 aggregate
sex          : T                    # Total
time         : 2014
value        : 27                   # 27% smoke boxed cigarettes, cigars,
                                    # cigarillos or a pipe

geo          : EU27_2020
sex          : M                    # Males
time         : 2014
value        : 32

geo          : EU27_2020
sex          : F                    # Females
time         : 2014
value        : 22

geo          : DE                   # Germany
sex          : T
time         : 2014
value        : 27

geo          : UK                   # United Kingdom
sex          : M
time         : 2014
value        : 23

Five rows, three facts. Across the EU-27 in 2014, 27% of the 15-plus population smoked - split 32% of men against 22% of women, a ten-point gap that is itself the headline for anyone forecasting combustible volumes by segment. Germany sat exactly on the EU average that year, which makes it a useful calibration row. UK men landed at 23%, nine points under the EU male figure, the kind of divergence that separates a European story from a British one.

Notice that every number arrives with its whole identity attached - entity code, sex, year, unit - so the panel assembles with a group-by rather than a parsing job. Your sample returns this exact dictionary scoped to the countries, sexes and survey waves you name; the schema in the sample is the schema that ships.

Seven fields carry the entire indicator, every one of them verified against the published column definition. Two of them are constants worth understanding rather than ignoring: age is pinned to Y_GE15, meaning the 15-plus population, and unit is pinned to PC_POP, percentage of total population. Constants are a design decision, not missing metadata - they guarantee that any two values you compare were produced on identical terms.

sex is the only live breakdown in the published dimension set - Total, Males, Females - and that constraint is the honest description of what this indicator offers: prevalence by sex, nothing finer. geo carries the full entity code set from EU aggregates down to member states and EFTA countries, time holds the survey-wave year, freq is always annual, and value is the percentage itself.

Additional fields on request: there is no education, income or age-band dimension in this table - if your model needs those cuts, say so at scoping and the companion sources that carry them get enumerated and defined against live records when your sample is cut, which is also where column naming locks for your pipeline.

• Geo: 35 entities under one code set - the EU-27 aggregate, euro-area EA20/EA21 aggregates, a separately maintained long-run EU series, all 27 EU member states, plus Iceland, Norway, Switzerland and the United Kingdom. Because historical UK rows survive Brexit inside the same scheme, a pre-2021 versus post-2021 comparison needs no reconciliation layer.
• Temporal: seven survey waves - 2006, 2009, 2012, 2014, 2017, 2020, 2023. The series is intermittent by construction: observations exist where Eurobarometer ran, and the gaps are honest survey cadence rather than missing data. The current table revision dates to January 2026, so the 2023 wave is the freshest reading.
• Granularity: one observation per geography x sex x survey year, ages 15+, expressed as a percentage of population. Roughly 620 observations in total - small enough to hold in your head, structured enough to join onto anything with a country column.

Set against the wider Datadory catalog - where the average quality score across all 1,744 datasets is 7.81 - this slice scores 9/10, carried by verified field documentation and a definition stable enough to span seventeen years of public-health change.

API, files, or your warehouse. Daily, weekly, or hourly.

Pick the channel your stack already speaks and set the cadence to match the decision being fed. Six hundred and twenty observations is a payload that fits anywhere - a dashboard widget pulling hourly, overnight files loading into BI, or a direct landing in Snowflake, BigQuery or Redshift beside your sales and excise tables. Cadence changes are a settings conversation, not a re-integration project.

Every delivery ships with the field dictionary above unchanged and sample rows for validation, one row per observation. Name the countries, sexes and waves when you request the sample; it comes back cut to that exact shape before any commitment.

Ranked by how directly one row settles the day job:

1. Public-health and SDG reporting teams. This is the official prevalence series behind SDG 3.4 non-communicable-disease monitoring in the European Statistical System - briefing tables that match the arithmetic Brussels itself reasons over.
2. Tobacco strategists and category planners. A contracting legal consumer base, sized country-by-country and sex-by-sex - the demand-side denominator underneath volume and excise models, delivered on a single definition.
3. Investors and quant researchers. Prevalence as a slow-moving regulatory-risk input for European cigarette exposure: seven waves is deep enough for trend features and sparse enough to force honesty about interpolation.
4. Insurers and actuaries. Self-reported smoking status by country and sex, feeding mortality and morbidity assumptions at portfolio grain rather than continent-wide averages.
5. Journalists and academics. Citable official percentages whose cross-country comparability was enforced upstream, so a quote survives scrutiny of its denominator.

The modeled alternatives trade self-report fidelity for breadth: WHO Global Health Observatory tobacco collection adds MPOWER policy indicators, IHME GHDx GBD 2019 smoking prevalence stretches to 204 countries, and the World Bank tobacco smoking indicators API covers the globe on modeled adult rates. Survey-grounded European depth and modeled global breadth are complements, not substitutes.

Investors and quants get a clean, officially defined prevalence panel to regress against tobacco-sector revenue and regulation - small data, but unimpeachable data. Data scientists get enum-coded dimensions and numeric values ready for pandas in minutes: no unit strings to strip, no country spellings to fuzzy-match. Market researchers and consultants get citable percentages for sizing decks, identical in definition across all 35 entities. Journalists and academics get the version of the prevalence numbers that public institutions themselves publish and defend. Across all of them the constant holds: one narrow indicator, measured identically, seventeen years deep.

Three things, stated up front. First, these are aggregate percentages, not microdata - there are no individual respondent records here, so anything requiring person-level analysis needs a different instrument. Second, the timeline is a wave grid, not a continuous series - tell us whether you want it as-is or bridged against modeled companions for annual continuity, because the two choices produce different panels. Third, sex is the only breakdown; age-band, education or income cuts inside Europe become a companion-source conversation we resolve at scoping rather than a surprise after delivery. Start at Get a sample of this dataset.

Notes that pair well with this page:

• The tobacco data hub collects every dataset in the industry, this one included.
• Best-of ranking: tobacco datasets puts this indicator in context against the alternatives.
• Policy counterpart: WHO Global Health Observatory tobacco collection - the MPOWER policy measures that act on the behavior this indicator tracks.
• Global extension: IHME GHDx GBD 2019 smoking prevalence - modeled smoker counts and cigarette-equivalents for 204 countries.
• Market view: Our World in Data - Smoking - curated prevalence, sales and attributable-death series built for communication.
• US angle: US TTB Tobacco Statistics - removals, production and tax-relevant volumes for the American market.
• Field guide: the tobacco data guide walks the full industry dataset stack.