Reading the 2022 Romanian Religion Census Without Getting Misled by the Headlines
The 2022 Romanian census included religion as a voluntary question for the first time in decades, and the results came out with enough noise that a lot of people drew conclusions that don't actually hold up. I spent about three weeks going through the raw tables from INS (Institutul Național de Statistică), cross-referencing them with the 2011 data, and talking to a few people who actually work in demographic research. What follows is how to read these numbers without tripping over the obvious traps. The headline number everyone quotes is that around 81% of respondents identified as Orthodox. That sounds dramatic if you've never looked at Romanian census methodology before, but it's not actually a surprise once you understand what the question asked and how it was administered. The key detail that most summaries skip is that the census listed specific denominations rather than leaving it open-ended. Respondents chose from a predefined list that included Romanian Orthodox, Greek Catholic, Roman Catholic, Baptist, Pentecostal, Adventist, Muslim, and a handful of smaller groups. "No religion" and "prefer not to say" were also options, but the framing of the question still pushed people toward identifying with an institutional group. Here is the part that catches people out. In 2011, the census asked about religion differently, and the participation rate was higher because it was treated as a more central demographic marker. By 2022, the voluntary framing meant that a significant number of people simply didn't answer the question at all. INS reported roughly 4.5 million people who skipped the religious affiliation question. When you divide the Orthodox percentage by the total population including non-respondents, the actual share drops closer to 60%. Both numbers are technically correct depending on whether you normalize for non-response. This matters because policy people and journalists almost always use the normalized figure without mentioning the adjustment.
One practical problem I ran into when trying to reconcile the 2022 data with the 2011 numbers was that INS published results at the county level but with different threshold rules for small communes. In some rural districts, if fewer than 200 people responded to the religion question, the data was suppressed or rounded in a way that made year-over-year comparison impossible. I worked around this by pulling the raw commune-level tables from the INS portal and manually reconstructing the suppressed entries using the 2011 baseline adjusted for population change from the civil registry. It took about two days of spreadsheet work but it was the only way to get a consistent picture at the local level. The Greek Catholic numbers are worth looking at separately from the main narrative. They dropped from roughly 0.8% in 2011 to about 0.5% in 2022. The immediate assumption is decline, but the reality is more about category salience. Many Greek Catholics in Maramureș and Satu Mare live in communities where the distinction between Greek Catholic and Orthodox blurs in practice, and when faced with a census checkbox, some respondents defaulted to the majority category. This isn't unique to Romania. Similar patterns show up in Irish census data around Catholic identification and in Polish data around religious practice versus nominal affiliation. Another counter-intuitive finding is that the Protestant categories, particularly Baptists and Pentecostals, showed modest growth rather than the flatline that some demographic models predicted. The combined Protestant share rose from about 0.9% to roughly 1.2% between the two census rounds. The growth is concentrated in urban centers and among younger cohorts, which makes sense given the migration patterns from rural areas to cities like Bucharest, Cluj, and Timișoara. But here's the nuance that trip people up: much of the apparent growth is cohort replacement, not conversion. Younger Romanians moving into cities are more likely to report a minority denomination than their rural parents did, partly because urban church networks are more visible and partly because the census question lands differently when you're no longer surrounded by a monoreligious community.
The Muslim population held steady at around 0.3%, which aligns with the Turkish and Tatar communities in Constanța and Tulcea. This number is often misunderstood as stagnation when it's really just a reflection of the fact that Romania doesn't have significant Muslim immigration compared to Western Europe. The census question doesn't capture undocumented residents or people who avoid answering for reasons unrelated to genuine non-affiliation. Now for the limitations, because this data has real blind spots. The biggest one is that religion in Romania is deeply entangled with national identity, so a lot of people who are culturally indifferent still check Orthodox because it's the default box for "Romanian." This means the census measures cultural identity more than belief or practice. If you want to know how many people actually attend services, this dataset won't tell you. The 2022 census didn't include a question on religious practice, only affiliation, so any analysis that conflates the two is working with incomplete information. A second limitation is the administrative lag. INS data release happens in phases, and the full methodological notes and weight adjustments weren't fully published until well after the initial results. If you're doing academic work with this data, you should plan on revising your numbers once the final weights come out. I've seen published papers that got corrected after the revised tables arrived, usually because the initial release used a different imputation method for missing values.
Get the Full Details
If you're trying to use this data for something practical, like planning outreach or understanding regional demographics, here's what I'd suggest. Start with the county-level tables from INS, download the raw CSV files rather than relying on summary PDFs, and always calculate both the normalized and unnormalized percentages. Cross-reference with the 2011 census using the same county boundaries, but be aware that some communes were merged or reclassified between the two rounds. The Nomenclature of Territorial Units for Statistics (NUTS) changes are minor but they do matter if you're doing granular analysis. For the non-respondent adjustment, I found that applying a simple hot-deck imputation based on age, gender, and county from the 2011 data gets you reasonably close to what INS would have produced, but it's not perfect. The systematic bias tends to run slightly toward higher reported Orthodoxy because non-respondents skew older and rural, and those populations have higher nominal Orthodox identification. If you need precision, the alternative is to supplement the census with the European Social Survey Romania module, which includes a religious affiliation question administered independently and can help calibrate the census undercount. The takeaway isn't that the census is useless. It's that it measures something narrower than people assume. It tells you about cultural and ethnic identification with religious groups, not theological commitment or spiritual experience. The "surprising trends" that make headlines are usually artifacts of question design, non-response patterns, or category confusion rather than genuine shifts in belief. Understanding that difference is what separates a useful analysis from a story that sounds good but falls apart under scrutiny.
