Short answer: mostly in September and October. By the end of October nearly every fish tagged in a Massachusetts bay had gone, and big Chesapeake fish summering in Massachusetts and Cape Cod bays started south in November.
That comes from acoustic tags, not almanacs. In a study of 159 tagged adults followed from 2008 to 2012, striped bass were in Massachusetts waters north of Cape Cod every year from May to November. The studies that tracked fish through a Massachusetts summer put the exits in a tighter band:
| Where | Tagged fish | Years | When they left |
|---|---|---|---|
| Plum Island Estuary, North Shore | 60, 13–25 in | 2005–06 | every one gone by Oct 31 |
| Plum Island Estuary | 59, 17–25 in | 2015 | 65% still heard in Sept, 13% in Oct; 58 of 59 gone by Nov 1 |
| Plymouth–Duxbury Bay, South Shore | 35, 15–32 in | 2011 | 7 left in Sept, 25 in Oct |
| Massachusetts and Cape Cod bays | Chesapeake fish over 31 in | 2014–18 | there June–Oct; south Nov–Jan in 2014 |
| Massachusetts waters | Hudson River fish, 27 in and up | 2016–18 | stayed until late Sept |
Two cautions travel with that table. Two of those receiver arrays came out of the water in the last week of October, so the tail may run a little later than it shows. And none of it is a date. At Plum Island the fish "tended to trickle out", in the researcher's words: not as groups, and not on the same day.
What tells them to go
The most direct answer comes from the Plymouth–Duxbury study. Hollema and colleagues followed 35 fish on 34 receivers, with 13 temperature loggers in the bay, and asked what predicted whether a fish was still there each week. Both things they tested did: water temperature and day length.
The temperature model's halfway mark, where a tagged fish was as likely gone as still there, came at a weekly bay average of 16.8 °C, about 62 °F. The day-length model's came at 11.6 hours of daylight. About 70 percent of the fish left in the last two weeks of October, after the bay dropped below that temperature.
The catch is in the methods. Temperature and daylight fell together so closely that the authors had to model them separately, so the study cannot say which one the fish were reading.
Off Maryland, where fish tagged in Massachusetts and Maine turn up on the way south, Rothermel and colleagues found that the temperature trend across the wider region over the previous weeks predicted when fish showed up better than the temperature where they were swimming. They think the timing is set back on the summer feeding grounds.
That is as far as the evidence goes: cooling water and shortening days, together, over weeks. We could find no study measuring whether a warm autumn holds the fish in New England longer.
Why one number does not travel
62 °F is a Duxbury Bay number, and it breaks as soon as you carry it north. Here is when the week-long average water temperature settled at or below 62 °F at NOAA stations along the coast, 2020–25 (the two buoys, 2020–24):
| Station | Date range across the years |
|---|---|
| Portland, Maine | Aug 29 – Oct 4, though July averaged 60–63 °F |
| Boston, offshore buoy | Sep 27 – Oct 21 |
| Cape Cod Bay buoy | Oct 5 – Oct 23 |
| Nantucket | Oct 5 – Oct 24 |
| Woods Hole, Newport, New London | Oct 16 – Oct 31 |
These are water dates, not fish dates. Two things still stand out. At Portland, 62 °F is ordinary July water, so the Duxbury number cannot be a universal switch. And south and west of the Cape the water holds its heat longer, while at any one station the date wandered by about two weeks or more from year to year. The late years were late everywhere: 2021 was the latest, or tied for latest, at every station.
We found no published tag study that pins down the fall departure from Maine, New Hampshire, the Islands or Rhode Island the way these do for Massachusetts bays. Treat any confident date for those waters as a guess.
The fall run is the exit
Before leaving, the tagged fish changed how they used their water. In Duxbury Bay they ranged over more of the bay as it cooled, and their weekly ranges peaked in the last two weeks of October. At Plum Island they had nearly quit the upper river and creeks by September and spent more time in the lower sound, toward the way out. North of the Cape, adults came and went through the Cape Cod Canal and along the outer Cape.
Then they head south, slowly. Hudson River fish that summered in Massachusetts averaged about 4 miles a day southbound, against 6 on the way north, and reached Virginia in early to mid January. For big Chesapeake fish, the trip north was packed into about 60 days of spring; the trip south spread over about 90.
So the fall run on your stretch of coast is a procession: your summer fish leaving, and fish from farther north passing through on the same road. It is not a new wave arriving for the season. Where they spend the winter, and whether every fish goes, is its own story.
What to do with it
Let the calendar start the watch and the water run it. September is when to start paying attention. After that, the week-long trend of your own water is the part that changes from year to year, and the part the research keeps pointing to.
Do not borrow a number. Duxbury's 62 °F describes one bay in one year. At Portland it describes July.
Late in the year, think about the ways out. At Plum Island the fish moved down toward the mouth before they left, and north of the Cape the exits were the Canal and the outer Cape. The canal's current, turn by turn, is on the Cape Cod Canal current page.
When your water is done, look south and west. The water there reaches the same mark later, and the fish leaving yours have to pass through it.
How StripedLines uses this
StripedLines is built for striped bass. It will not turn Duxbury Bay's curve into a countdown for your water. With no receiver in your water, nothing can tell when the fish have left it; your water's temperature and your own log are the better guide.