Loto 6/49 · official archive, 4 January 1998 – 1 October 2026
Loto 6/49: the draws are close to random, the numbers people play are not
All 2,376 draws in the Romanian Lottery's archive, put to the test: how often each ball came up, “overdue” numbers, rare patterns and the number of winning entries. The balls show one small departure: small numbers come up slightly more often than they should. The counts of winning tickets show that the numbers people play are not chosen at random. Choose six numbers to see how close the draws came and how often numbers like yours are played.
- The ballsThe 49 balls came up almost equally often. The one departure, a small one: small numbers come up slightly more often, mostly since 2015 (average sum 146.3, against an expected 150).
- Overdue numbersA ball that had been missing for 32 draws or more came up in the next draw 13.2% of the time. Any ball's chance is 12.2%, and the difference fits within chance.
- The playersDraws whose six numbers were all 31 or below had 18% more 4/6 wins than a typical draw, for the same number of 3/6 wins.
Data up to 1 October 2026
Choose six numbersYour combination
Choose six numbers and see how close the draws came
Every combination has the same chance: one in 13,983,816. Choose six numbers and compare them with all 2,376 draws in the archive, up to 1 October 2026.
The interactive choice is unavailable during this visit. The numbers, explanation and sources remain available below.
Draws sit on rings according to how many numbers they share with your selection.
Closest draw
0/6
Choose six numbers to start the comparison.
Closest draws
How the 2,376 draws were distributed
The solid bar shows the archive. The thin marker shows the mathematical mean for the same number of draws.
How often numbers like yours are played
This shows how often numbers like yours are played. It does not change your chance: one in 13,983,816. Combinations with a pattern, such as 1–2–3–4–5–6, may be played far more often, and that is not visible here. How this was measured
The scale of the game
The same chance for every combination
The combination 1–2–3–4–5–6 has the same probability as any other: one chance in 13,983,816 at every draw. The archive describes the past and does not change that fraction.
Closeness has seven exact steps
For any six-number choice, all 13,983,816 possible combinations fall into exact groups according to how many numbers they share with it.
Why no combination has repeated in 2,376 draws
2,376 draws form 2,821,500 pairs. With a one-in-13,983,816 chance for each pair, 0.2 repeats were expected, and the probability of seeing at least one by 1 October 2026 was 18%. “No repeat” is the most likely outcome.
At the pace of recent years (105 draws a year), the probability reaches 50% after 4,404 draws, around the year 2046. The chance that the next draw repeats one of the 2,376 combinations already drawn: 0.0170%.
The ball test
Are the draws random?
A fair lottery does not draw the balls perfectly evenly: some come up more often by pure chance. The question is whether the differences in the archive are larger than chance would give. Over 2,376 draws, each ball should come up about 291 times, with a typical spread of ±16.
How many draws each ball came up in
- 1283
- 2308
- 3290
- 4320
- 5334
- 6332
- 7294
- 8289
- 9303
- 10297
- 11275
- 12311
- 13304
- 14277
- 15285
- 16296
- 17305
- 18295
- 19283
- 20285
- 21286
- 22276
- 23305
- 24292
- 25316
- 26295
- 27303
- 28285
- 29273
- 30295
- 31257
- 32298
- 33302
- 34280
- 35261
- 36316
- 37279
- 38280
- 39298
- 40275
- 41290
- 42261
- 43288
- 44296
- 45284
- 46286
- 47267
- 48290
- 49256
A magenta ring marks a ball outside the range that holds 95% of what chance gives (260–322).
4 of 49 balls fall outside that range; a fair lottery gives 2 or 3. The overall deviation (χ² = 49.1) is matched or exceeded by 24% of simulated archives of the same size.
In a fair lottery, χ² falls between 30 and 58 in 90% of cases, and the gap between the most and the least frequent ball between 56 and 91.
The table of all 49 balls
| Ball | Draws | Against the expected mean |
|---|---|---|
| 1 | 283 | −2.7% |
| 2 | 308 | +5.9% |
| 3 | 290 | −0.3% |
| 4 | 320 | +10.0% |
| 5 | 334 | +14.8% |
| 6 | 332 | +14.1% |
| 7 | 294 | +1.1% |
| 8 | 289 | −0.7% |
| 9 | 303 | +4.1% |
| 10 | 297 | +2.1% |
| 11 | 275 | −5.5% |
| 12 | 311 | +6.9% |
| 13 | 304 | +4.5% |
| 14 | 277 | −4.8% |
| 15 | 285 | −2.0% |
| 16 | 296 | +1.7% |
| 17 | 305 | +4.8% |
| 18 | 295 | +1.4% |
| 19 | 283 | −2.7% |
| 20 | 285 | −2.0% |
| 21 | 286 | −1.7% |
| 22 | 276 | −5.1% |
| 23 | 305 | +4.8% |
| 24 | 292 | +0.4% |
| 25 | 316 | +8.6% |
| 26 | 295 | +1.4% |
| 27 | 303 | +4.1% |
| 28 | 285 | −2.0% |
| 29 | 273 | −6.2% |
| 30 | 295 | +1.4% |
| 31 | 257 | −11.7% |
| 32 | 298 | +2.4% |
| 33 | 302 | +3.8% |
| 34 | 280 | −3.8% |
| 35 | 261 | −10.3% |
| 36 | 316 | +8.6% |
| 37 | 279 | −4.1% |
| 38 | 280 | −3.8% |
| 39 | 298 | +2.4% |
| 40 | 275 | −5.5% |
| 41 | 290 | −0.3% |
| 42 | 261 | −10.3% |
| 43 | 288 | −1.0% |
| 44 | 296 | +1.7% |
| 45 | 284 | −2.4% |
| 46 | 286 | −1.7% |
| 47 | 267 | −8.2% |
| 48 | 290 | −0.3% |
| 49 | 256 | −12.0% |
Do small numbers come up more often?
The six numbers drawn should add up to 150 on average. In the archive, the average sum is 147.7. In a fair lottery, a departure this large from 150, in either direction, turns up in 0.05% of cases (0.1% of simulated archives). After correcting for the 15 tests on the page, the p-value is 0.008.
The band shows where 95% of the averages of a fair lottery fall; the dashed line marks 150.
The difference is small: in 2015–2026, numbers up to 24 made up 50.6% of the balls drawn, against an expected 49.0%. The periods in the table were chosen after seeing the result for the whole archive; before 2015, the averages sit inside the band. The archive does not say why: it could be an effect of the machine or the balls, or chance.
Thirteen rare patterns, counted
The probability of each pattern in one draw is calculated exactly from all 13,983,816 combinations. The table sets it against the real number of draws.
| Pattern | Chance in one draw | Expected | In the archive | 95% interval |
|---|---|---|---|---|
| At least one number from the previous draw | 56.4% | 1,340 | 1,312 | 1,292–1,387 (inside) |
| Two neighbouring numbers (such as 17 and 18) | 49.5% | 1,177 | 1,177 | 1,129–1,224 (inside) |
| Three numbers ending in the same digit (such as 7, 17, 37) | 9.22% | 219 | 222 | 192–247 (inside) |
| The six numbers add up to 100 or less | 6.69% | 159 | 164 | 135–183 (inside) |
| The six numbers add up to 200 or more | 6.69% | 159 | 132 | 135–183 (outside) |
| Four numbers from the same block of ten (such as 20–29) | 5.48% | 130 | 134 | 109–152 (inside) |
| All six numbers up to 31 (days of the month) | 5.27% | 125 | 141 | 104–147 (inside) |
| Three numbers in a row (such as 22, 23, 24) | 4.77% | 113 | 112 | 93–134 (inside) |
| All six numbers odd | 1.27% | 30 | 34 | 20–41 (inside) |
| All six numbers between 25 and 49 | 1.27% | 30 | 21 | 20–41 (inside) |
| All six numbers even | 0.96% | 23 | 18 | 14–33 (inside) |
| All six numbers between 1 and 24 | 0.96% | 23 | 30 | 14–33 (inside) |
| Six numbers in arithmetic progression (such as 5, 10, 15, 20, 25, 30) | 1 / 64,740 | 0.04 | 0 | 0–1 (inside) |
One pattern falls outside its interval: “The six numbers add up to 200 or more”: 132 against 135–183 expected. Across 13 tests, chance gives 0.65 on average. After correcting for multiple tests (Holm), the smallest p-value is 0.34, so no significant departure.
What a rigged lottery would have shown: with 2,376 draws, a ball favoured by about 23% would have stood out in 80% of cases, at a threshold corrected for the 49 balls. Differences of a few percent stay below what any archive of this size can show.
Overdue numbers
Are there “overdue” numbers?
Many people believe a ball that has not come up for a long time must come up. For every ball and every draw I counted how many draws had passed since its last appearance, then whether it came up. The chance is 12.2% at every draw, however long the gap.
The magenta line marks 12.2%. The band shows where 95% of results fall when the order of the draws is shuffled at random (1,000 shuffles).
One category falls slightly outside the band (2–3: 11.80% against a band of 11.84%–12.59%), below the line rather than above it, so against the idea of being “due”. With 7 categories, a fair archive has one outside its band in 30% of cases. The category with the longest gaps stands at 13.2%, but its band is wide (10.9%–13.8%) because it holds only 1,663 cases.
The most frequent balls
- 5334 times
- 6332 times
- 4320 times
- 25316 times
- 36316 times
- 12311 times
The rarest balls
- 49256 times
- 31257 times
- 42261 times
- 35261 times
- 47267 times
- 29273 times
Missing the longest after the 1 October 2026 draw
- 3843 draws
- 2942 draws
- 4728 draws
- 2626 draws
- 3520 draws
- 1315 draws
- 3615 draws
- 4515 draws
The longest gap in the archive was 81 draws. No gap changes the chance of the next draw: 6 in 49 for every ball.
The players
The numbers played are not chosen at random, and the wins show it
For every draw, the archive says how many wins there were at 3, 4, 5 and 6 numbers matched. A win is a winning ticket or system line, not a player. Wins at 3/6 give an approximate measure of how many tickets were played; their number also depends on the combinations played and on those drawn. If all tickets were chosen independently and at random, the expected ratio of 4/6 wins to 3/6 wins would be the same at every draw.
The observed ratio changes with the numbers drawn, though: the more numbers above 31, the fewer 4/6 wins. Players choosing calendar dates, which stop at 31, is one possible explanation; the archive does not say why players choose their numbers.
4/6 wins, for the same number of 3/6 wins, by how many drawn numbers were above 31
The bar shows the difference from a typical draw; the band, the 95% interval.
Across 1,631 regular draws (2 May 2010 – 1 October 2026), each extra drawn number above 31 brings a change of −5.8% in 4/6 wins (95% interval: −6.4% to −5.2%). A typical draw has about 509 4/6 wins.
The effect holds in every variant tried: from 2012 only, up to 2018 and from 2019 on, without 2020–2021, with the size of the jackpot taken into account (between −6.1% and −5.6% per number). At 5/6 against 3/6 the estimated association is larger, −10.5% (95% interval: −13.2% to −7.6%). If I redistribute the combinations among the draws at random (2,000 redistributions), the slope stays between −1.3% and +1.3%, so no shuffle comes near the real one.
The effect of each number on its own
- 1−1.0%
- 2+1.5%
- 3+4.7%
- 4+1.0%
- 5+4.1%
- 6+2.3%
- 7+6.3%
- 8+4.3%
- 9+6.8%
- 10+3.8%
- 11+2.4%
- 12+4.3%
- 13+7.5%
- 14+1.6%
- 15+0.7%
- 16+1.4%
- 17+5.9%
- 18+1.4%
- 19+2.3%
- 20−1.7%
- 21−1.5%
- 22−0.1%
- 23+3.6%
- 24+3.6%
- 25+3.4%
- 26+2.6%
- 27+3.5%
- 28+0.2%
- 29+0.8%
- 30−1.5%
- 31−4.8%
- 32−4.9%
- 33+0.9%
- 34−3.2%
- 35−2.3%
- 36−1.0%
- 37−0.1%
- 38−1.7%
- 39−4.3%
- 40−5.5%
- 41−5.8%
- 42−5.0%
- 43−5.3%
- 44−3.0%
- 45−6.3%
- 46−4.0%
- 47−4.8%
- 48−4.5%
- 49−5.3%
more wins fewer wins
Each estimate carries an uncertainty of about ±2 percentage points, so the numbers are not ranked against each other. The average of the estimates is +3.4% for 1–12, +1.5% for 13–31 and −3.7% for 32–49: the difference between the groups is clear, the difference between neighbouring numbers is not.
Who shares the jackpot?
Of 198 jackpots won at 6/6 in regular draws, 137 had a single win and 61 had several wins (the archive does not say whether they belong to different players); 53 of those are from 1998–2001. In that period, random choice would have given 1.1 wins at 6/6 per draw on average. From 2010 on, the expectation is 0.038; the jackpot was won in 60 of 1,664 draws, and one had several wins, on 10 October 2010.
| Period | Draws | Jackpots won | Expected with independent tickets | With several wins (observed / expected) |
|---|---|---|---|---|
| 1998–2001 | 207 | 94 | 110 ± 6 | 53 / 58.6 |
| 2002–2009 | 436 | 44 | 45 ± 6 | 7 / 4.3 |
| 2010–2026 | 1,664 | 60 | 61 ± 8 | 1 / 1.6 |
The expectations assume independent tickets chosen at random, with volume estimated from the 4/6 wins. The ± is the typical spread of the number of jackpots and leaves out the uncertainty of the volume estimate.
In 1998–2001, fewer jackpots were won than ticket volume would have given (94 against 110), but the total number of wins at 6/6 was higher (262 against 230). That is compatible with wins concentrating on some combinations, but the benchmark depends on the estimated volume and on tickets being independent.
The jackpots with the most wins
- 12131825333818 wins
- 581317242815 wins
- 5172324343511 wins
- 13182632384610 wins
- 1317293339417 wins
What it means if you play
The chance of winning does not depend on the numbers chosen: one in 13,983,816. In the data from 2010 on, a jackpot has rarely had several wins. Draws with more numbers above 31 had fewer 4/6 wins relative to 3/6 wins; the analysis does not estimate how many other wins a particular choice would have shared a jackpot with.
How to read the numbers
The data, the tests and the limits
The data
Loto 6/49 results published by the Romanian Lottery, from 4 January 1998 to 1 October 2026: 2,376 draws, of which 67 were special draws held on the same day as a regular one. Special draws are included in the tests on numbers. The analysis of wins leaves them out, because each has its own prize fund, and it also leaves out the two draws of 1 December 2019, the same day, which carry no label. The numbers of each draw are sorted. The data were retrieved on 2 October 2026 and checked against the official list of latest results.
The tests
Pattern probabilities are calculated exactly from all 13,983,816 combinations. The average sum is compared with the exact value 150 and with 2,000 simulated archives. Ball frequencies are compared with 4,000 simulated archives of the same size, and the gaps with 1,000 shuffles of the order of the draws. The simulations use seed 49 and are simulations, not official data; they are labelled wherever they are read.
The wins
The model counts 4/6 wins with 3/6 wins as the measure of volume and a separate effect for each year (Poisson regression, errors widened for variation beyond Poisson). It uses 1,631 regular draws from the period in which the archive publishes 3/6 wins.
The limits
The tests can rule out large departures, not departures of a few percent. Wins show which numbers are played on average; they say nothing about an individual player's ticket and do not measure combinations with a pattern, such as 1–2–3–4–5–6, which may be played far more often. Jackpot amounts before June 2005 are in old lei and enter no comparison.
Sources
- Romanian Lottery, Loto 6/49 results by month
- Romanian Lottery, simple-entry rule
- Romanian Lottery, latest 6/49 results