What a Bear Market Is and How Long They Tend to Last

A bear market is a fall of 20% or more in a broad stock index; the SEC's Investor.gov says it generally happens over at least a two-month period. Measured on S&P 500 closing prices from 1957 to October 2026, 11 declines met the 20% test, lasting from about 1 to 30 months from record high to low, with a median near 9 months and a median depth of 33.5%. Getting back to the old high took a median of about 15 months after the low, and between 3 and 70 months across the 11.

Key takeaways

• Investor.gov's definition has two parts, a depth of 20% or more in a broad index and a time span of at least two months; two of the 11 declines (1987 and 2020) reached 20% in under two months.
• Peak-to-low length ranged from 1.1 to 30.5 months (median 9.3), and depth from 20.7% to 56.8% (median 33.5%), so no single "typical" bear market exists.
• Crossing the 20% line did not mark the bottom: from the first close 20% down, the further fall ranged from 0.4% to 45.6%, with a median of 8.2%.
• Seven of the 11 bear markets overlapped an official recession; four (1961–62, 1966, 1987, 2022) did not, and three recessions (1960–61, 1980, 1990–91) had no 20% bear market.
• A larger fall needs a disproportionately larger gain to recover: a 33.5% loss needs about a 50% gain to get back to the old level.

What the 20% rule says

The SEC's investor education site defines a bear market as a time when stock prices are declining and market sentiment is pessimistic. It adds that a bear market generally occurs when a broad market index falls by 20% or more over at least a two-month period. The wording says "generally," and it does not name an index, a price basis (closing or intraday) or a starting point.

This guide fills those gaps in one way, and other analysts fill them differently, which is why published counts and dates can vary. The calculation uses the S&P 500, measures from the highest closing level reached so far (the record close) and counts a bear market whenever a later close is at least 20% below it, whatever the time span. The decline ends at the lowest close before a new record is set.

For 1957 the starting point is the highest close since the index's March 1957 launch; Yahoo's series shows a higher close in August 1956, from before the 500-stock index existed.

decline = close on a later day / record close before it - 1
recovery gain needed = 1 / (1 - decline) - 1

decline              = fall from the record close, as a negative percentage
recovery gain needed = percentage gain from the low required to regain the record
                       (a 20% fall needs 25%, because 0.80 x 1.25 = 1.00)

That rule is not neutral at the margins. In 1990 the S&P 500 fell from 368.95 on July 16 to 295.46 on October 11, a decline of 19.9% on a closing basis, and it is not counted here. A conventional line splits near-identical events, so a decline of 18% or 22% differs less than the label suggests.

The 11 bear markets since 1957

The figures are this article's own calculation from daily S&P 500 closes in Yahoo Finance's ^GSPC history, from March 4, 1957 (the commonly reported start of the 500-stock index) to October 12, 2026. Yahoo is a commercial data service, not an official record. The St. Louis Fed's FRED SP500 series covers only about the last ten years; over the 2,512 days the two sources share, closes differed by at most 0.12%, but the earlier bear markets rest on Yahoo alone. Prices exclude dividends, so recovery dates with dividends reinvested would be the same or earlier.

The calculation found 23 declines of 10% or more from a record close in that period; the 11 below reached 20%. Months are day counts divided by 30.44.

Summarized across the 11:

What the table shows about length

The median bear market took about nine months to fall, but the mean is 12.5 months because five of the 11 (1968–70, 1973–74, 1980–82, 2000–02 and 2007–09) each ran 17 months or more. The shortest were 2020, at about one month, and 1957 and 1987, at about three months each. Recovery is even more skewed: the median is 15.2 months from the low, the mean 24.5, and the 1973 peak was not regained until July 1980, about 90 months after the high.

Reaching 20% did not mark the low

The two-month test is about how a 20% fall is judged, not how far prices go afterward. In seven of the 11 cases the index closed less than 10% below the first 20%-down close at its eventual low. In the other four (1968–70, 1973–74, 2000–02, 2007–09) it fell a further 19% to 46%. The median further fall was 8.2%.

The median time from the record close to the first 20% close was 6.6 months, and from that first 20% close to the low a further 3.8 months, ranging from one day (1957) to 18.9 months (2000–02, from March 2001 to October 2002).

A reader who sees a 20% headline has no way, from the number alone, to tell which group the decline belongs to.

The two-month wording and the fast declines

The first 20%-down close came 1.8 months after the record in 1987 (the October 19 crash) and 0.7 months after it in 2020. The 2020 decline fails "at least a two-month period" on any reading: the whole fall from peak to low took 1.1 months. The 1987 decline crossed 20% in under two months but kept falling for 3.3 months to a low 33.5% below the high, so whether it qualifies depends on which span is measured.

This guide counts both on the 20% test alone; when comparing published lists, check whether the source applies the time clause.

Bear markets and recessions

A bear market is a price event. A recession is a judgment about the economy made by the National Bureau of Economic Research (NBER), which looks at depth, spread across the economy and duration, and treats the peak month as the last month of an expansion, so a recession runs from the month after a business-cycle peak to the trough month. Its business cycle dating page explains the method, and its table of US business cycle peaks and troughs gives the dates. The NBER says it waits a number of months after a peak or trough has occurred before identifying it, so a recession cannot be confirmed at the time a bear market is unfolding.

Comparing the 11 bear markets with the 10 NBER recessions whose peak month was August 1957 or later gives an uneven overlap (dates below are the NBER peak and trough months):

• Overlapped a recession (7): 1957 (peak August 1957, trough April 1958), 1968–70 (December 1969, November 1970), 1973–74 (November 1973, March 1975), 1980–82 (July 1981, November 1982), 2000–02 (March 2001, November 2001), 2007–09 (December 2007, June 2009) and 2020 (February 2020, April 2020).
• No recession during the decline (4): 1961–62, 1966, 1987 and 2022.
• Recessions without a 20% bear market (3): 1960–61 (peak April 1960, trough February 1961), 1980 (January 1980, July 1980) and 1990–91 (July 1990, March 1991); the index fell 19.9% in 1990, as noted above.

Overlap in time is not proof of cause in either direction, and the order of events varied. In 1957 the index peaked on July 15, a month before the NBER's August business-cycle peak, and first closed 20% down on October 21. In 2000–02 the index low came on October 9, 2002, 11 months after the NBER's November 2001 trough. A bear market can occur without a recession, and a recession without a bear market; neither one is a reliable signal for the other in this record.

Why recovery usually takes longer

Percentage losses and gains are measured from different starting points. After a fall, the base is smaller, so a gain of the same percentage recovers less. The recovery formula above gives the gain needed from the low to regain the record close:

The asymmetry is arithmetic, not a market pattern. It sets the size of the gain needed; how long that gain takes depends on how fast prices rise. Recovery took longer than the fall in nine of the 11 cases. The exceptions were 1966 (7.9 months down, 6.9 back) and 1980–82 (20.4 months down, 2.7 back, a 37.2% gain in under three months).

The arithmetic is part of why the 2007–09 index needed 17.0 months to fall and 48.6 months to return to its old high: it had to more than double from its low (a gain of about 131%). A portfolio's own recovery also depends on what it held, how much was added or withdrawn during the decline and whether dividends were reinvested, none of which an index price series shows.

Frequently asked questions

How often do bear markets happen?

The 11 declines fall in the 69.6 years from March 1957 to October 12, 2026, about one every 6.3 years on average. The gaps were uneven: 12.6 years separated the 1987 and 2000 peaks, and under two years separated the 2020 and 2022 peaks. An average interval says little about when the next one begins.

Is the market in a bear market now?

Not on this test. The S&P 500 closed at a record 7,818.93 on October 6, 2026 and at 7,807.24 on October 12, 2026, 0.1% below that record. The October 12 close matches the FRED series, and the figures will be out of date once the index moves.

Does an intraday drop of 20% count?

Investor.gov does not specify a price basis, so definitions differ. This guide uses closing prices, which is why the 1990 decline of 19.9% is excluded. Another analyst using intraday lows could date some bear markets differently, and any two sources can disagree on starts and ends without either being wrong.

Why do stocks sometimes bottom before the economy does?

In this record they usually did: in six of the seven bear markets that overlapped a recession, the index low came before the NBER trough month. The 2007–09 low came on March 9, 2009, three months before the June 2009 trough, and the 2020 low came on March 23, a month before the April 2020 trough. The exception was 2000–02, when the low came 11 months after the November 2001 trough. Prices reflect expectations about the future while the NBER dates reflect measured activity, so the two can move at different times.

Do dividends change how long recovery takes?

They can only bring the date forward. The price series used here leaves out dividends, so the dates for regaining the old high are the latest case for an index fund that reinvested payouts. This article did not calculate the total-return version, so the size of the difference is not shown here.

Checklist for bear-market headlines

Forecasts of when a bear market starts or ends are not supported by the record above: the further fall after the first 20% close ranged from under 1% to nearly 46%, and the 20% reading itself looked the same in both groups. What a reader can do is check the claim before accepting it.

1. Identify the index: a decline in the S&P 500, a single sector or a narrow index is a different event.
2. Check the basis: closing or intraday prices, and from which record high the 20% is measured.
3. Check whether the figure is price only or includes dividends.
4. Compute the gain needed to recover with the formula above rather than relying on the label.
5. Treat any statement such as "this bear market will end in X months" as a guess unless it shows its method; the 11 durations ranged from about 1 to 30 months.

Sources

• Investor.gov (SEC), glossary: Bear market
• National Bureau of Economic Research, Business Cycle Dating
• NBER, US Business Cycle Expansions and Contractions
• Federal Reserve Bank of St. Louis, FRED: S&P 500 (SP500)
• Yahoo Finance, ^GSPC historical prices (secondary source, used for 1957 onward)