A whimsical proposal: Overnight Sleeper Services for Boston

This is a pie-in-the-sky idea that barely merits being called a “proposal”. Nevertheless, I believe it is thought-provoking and worth consdering as we imagine what we want Boston’s relationship to the rest of the country to look like in the new millennium.

Riverside presents… overnight sleeper services for Boston.

This piece was largely written in 2019 and reflects the specific schedules in place at that time; current schedules may differ in particulars, though are (to my knowledge) largely consistent overall.

What Do I Mean By “Overnight”?

For my purposes, I’m defining “overnight service” to mean “boarding in the evening in one city, sleeping aboard, and disembarking in a new city not long after waking.” Like a red-eye, you save on overnight accommodations and don’t waste daytime hours traveling. With a red-eye, you usually pay for those benefits at the cost of comfort and — usually — any decent quality of sleep. An overnight train service will usually have those same costs, although the comfort should be a bit higher, and the cost of a more comfortable seat (e.g. business class) will probably be lower.

But overnight trains have a couple of notable differences. First, even when just traveling in coach seats, you generally get a one-seat ride. For example, some red-eyes require a late-night or very early-morning transfer to reach your destination; even if you’re lucky enough to get comfortable on the first leg, you have to start all over on the second. Second, it is possible to get an actual bed and privacy on an overnight train that you basically cannot get on a plane without paying a four-digit sum.

Finally, depending on the particulars, there are certain journeys that can be accomplished relatively quickly through air travel, but which take just long enough, especially when airport overhead travel time is accounted for, that they can still cost you most of a day. Some of those journeys could instead be accomplished with overnight rail with less overall disruption to one’s week.

Additionally, there is the growing public awareness of the carbon impact of air travel, which may lead to renewed interest in alternate methods for medium-distance journeys.

Current Services

As it is, Boston currently has what might count as “one and half” overnight services — depending on how you count them.

Northeast Regional

Trains 65 and 67 leave Boston at 9:30PM. The arrival time to New York is really too early to count as “overnight”, since you still get there right in the middle of the night (2:30 AM). However, once you get to Philadelphia (4:30 AM), you just start to slip into “overnight” territory. Certainly journeys to Washington (6:30 AM) or Richmond (9:00 AM) would count (9 hours and 11.5 hours, respectively).

I’d say the Northeast Regional counts as somewhere between one-half and one overnight service. The biggest challenge is the lack sleeping accommodations on this train. Amtrak experimented with sleeper cars on this route in the last few years, but for now has not reinstated that particular aspect of the service.

It should be acknowledged that there are many transportation options on this corridor, meaning sleeper services face that much more competition. The timing of departures and arrivals means that true “overnight” journeys are limited to Boston/Providence to Washington/Baltimore — hardly an underserved market, especially given that BWI is a hub for Southwest.

That being said, I suggest that Amtrak should reinstate its sleeper service on this corridor. It’s simple and straightforward for passengers to understand, it provides a reasonable alternative to flying, and the Boston-Washington corridor is exactly the kind on which we should encourage mode shift from air to rail whenever possible; the daytime rail journey is just long enough that an overnight service could be significantly more convenient.

Lake Shore Limited

Train 449 leaves Boston at 12:50PM, combines with 49 at Albany from which it departs at 7:05PM, and arrives in Chicago the next day at 9:50AM. Certainly from Albany onward, this counts as an “overnight”, but adding on the 6 hours at the start from Boston is brutal. And, it’s worth adding, that 9:50 arrival in Chicago is definitely pushing the convenience level, as it definitely eats into your day (and that’s before any delays on this 1,000 mile journey). So, from a Boston perspective, this is also probably not a proper “overnight” service.

Unlike the Northeast Regional, the Lake Shore Limited does offer sleeper accommodations. A one-way ticket from Boston to Chicago for a roomette is roughly $400. By comparison, a flight from Boston to Chicago, plus a hotel in Downtown Chicago is between $200 and $370, depending on economy vs business, and whether you check bags. Plus add $40 to reach downtown from the airport by cab.

So, for a business traveler who can sacrifice half of her time in the office on Tuesday, the cost for an overnight Amtrak might be just comparable to flight + hotel if she needs to be in Downtown Chicago for a mid-day meeting on Wednesday. (Barely.) To be fair to Amtrak — if our traveler opted for air instead of rail, she would either need to chop off a few hours of her workday for a 5pm flight out of Logan, or would need to do a late (10pm CT) arrival into Chicago, probably not reaching hotel until close to 11pm CT/midnight ET. An Amtrak roomette won’t be as comfortable as a hotel room, but at least you can be in bed at a reasonable hour.

The other variable to note here: on average, Train 49 is 45 minutes late into Chicago, according to the Amtrak Status Maps Archive Database. A delayed flight will keep you from getting to bed, but won’t make you late for that mid-day meeting — a delayed train arriving that same morning might.

The Amtrak fact sheet on the Lake Shore Limited offers some interesting data points. First, a full 21% of journeys are simply between Albany and New York — part of the frequency layer cake that amounts to a train every 60-90 minutes between Albany and New York for most of the day (!). All but one of the remaining top 10 city pairs involve Chicago. Trips between Boston and Chicago only seem to account for 3.1% of ridership, though NYC-CHI only does a bit better at 6.9%. However, even among these smaller groups, Sleeper reservations still make up a minority — not even half, by the looks of it.

The low share of Sleeper reservations may in part be because Coach is much cheaper — the $400 trip above is $96 in the cheapest tier of Coach. Shockingly, that is still in fact more expensive than some flights (when booked in advance), though perhaps if you are able to avoid a cab fare it might be slightly cheaper.

Unlike the BOS-WAS corridor described above, BOS-CHI is probably a bit too long to be attractive for overnight sleeper travel.

But there are a couple of other data points I want to highlight from the LSL schedule and ridership, as I think it does inform what could work in Boston.

Lake Shore Limited from Chicago — a model for success?

I’d like to focus on this middle tier of city pairs:

3. Buffalo, NY – Chicago, IL 528 mi
4. Chicago, IL – Syracuse, NY 668 mi
5. Albany-Rensselaer, NY – Chicago, IL 818 mi
6. Chicago, IL – Rochester, NY 589 mi
7. Chicago, IL – Cleveland, OH 341 mi
8. Chicago, IL – Toledo, OH 234 mi

I think there is something very interesting here. Notice that the shorter-distance pairs (with Cleveland and with Toledo) are less patronized than the longer journeys to Rochester, Albany, Syracuse and Buffalo. Likewise, population is not particularly predictive: the Albany, Buffalo, and Rochester CSA’s are roughly 1.1, 1.2, and 1.1 million respectively, while Cleveland’s is at a whopping 3.6 million — the largest on this corridor aside from its endpoints, but wallowing down in 7th place on this list. Likewise, Syracuse at 730K handily trounces Toledo’s 830K, despite being more than twice as far to travel.

This dip in ridership makes perfect sense — once you look at the schedule:


Those scheduled times for Cleveland and Toledo are awful. In both cases, one direction is sorta okay — Cleveland to Albany leaves just before 6AM, and likewise for Toledo to Chicago — but the other direction requires you to be at a train station at 3 in the morning. That’s just terrible (as I’m sure Amtrak realizes.)

Contrast that with the brackets created by our more popular destinations. Going east, a 9:30PM CT departure from Chicago will get you to

  • Buffalo at 8:46AM
  • Rochester at 9:53AM
  • Syracuse at 11:23AM
  • Albany at 2:31PM

Likewise, departing from

  • Albany at 7:05PM
  • Syracuse at 9:52PM
  • Rochester at 11:12PM
  • Buffalo at 12:20AM

will get you to Chicago at 9:50AM CT.

Now, all of these are a bit on the late side — it would probably be nicer to leave Chicago earlier and be in Rochester in time for breakfast — and that Buffalo departure is rough, plus the Chicago arrival is a bit late in the morning — but all of them would allow you to get a reasonable night’s sleep.

All of this suggests that medium-distance overnight journeys of 10-13 hours might be a “sweet spot” wherein longer travel times are tolerated because they are convenient.

Finally, it’s worth noting that some 225,000 riders in 2018 took the Lake Shore Limited in Coach (not counting those ALB-NYC supercommuters). This is more than the 110,000 who took the daytime Adirondack, and is comparable with the overnight Capitol Limited, daytime Palmetto, and overnight City of New Orleans, which travel similar distances. So, even though it may be a bit of a drag, sleeping in a Coach seat in order to travel ~800 miles overnight definitely seems to be done.

(By the way — of those last three, the Capitol Limited sees by far the most Sleeper passengers, and most of those are Chicago-Washington, with Chicago-Pittsburgh coming in a reasonable 2nd place. WAS-CHI is 4:05PM-8:45AM and CHI-WAS is 6:40PM-1:05PM — though a long journey, those timings are great, especially the westbound. The Pittsburgh times are a little rougher — depart at midnight going west, arrive at 5AM coming east, but it’s not bad, especially with the time change taken into account. I’d wager that the higher share of Sleeper passengers on this route is explained in part by the maximally convenient times.)

All of which is to say: Among current medium-distance overnight riders, convenient “traveling time brackets” seem to play a non-trivial role in encouraging and discouraging ridership.

Proposed Services

So where does this leave Boston?

With the above in mind, I propose four corridors for overnight service from Boston:

  • Boston-Albany-Montreal
  • Boston-Albany-Buffalo(-Toronto)
  • Boston-Albany-Buffalo-Cleveland and onward
  • Portland-Boston-Washington


By splicing together the Lake Shore Limited and Adirondack schedules — and assuming that Gare Centrale’s Customs Facility finishes up soon — we arrive at a travel time of about 12 hours via Albany. Now, for daytime travel, this would be pretty rough — not at all competitive against driving. However, as an overnight service, it may just fall into our sweet spot.

An 7:30pm departure from South Station would reach

  • Worcester at 8:45
  • Springfield at 10:00
  • Pittsfield at 11:15
  • Albany at 12:30

before arriving in Montreal at about 7:30am the next morning. And that assumes no track upgrades in Massachusetts, and assumes all currently scheduled Adirondack stops (except for St Lambert) are made. You could probably pick up some time by trimming the Albany layover (23 minutes in this exercise) and by dropping most of the Adirondack stops. Assume an hour within the station for customs, and you’re out having breakfast in the Old City by 9am.

Note that those stop timings correspond very closely with the pattern we saw on the westbound Lake Shore Limited with Albany through Buffalo. That’s a sweet spot that could work.

Especially if positioned as a weekend getaway train, a la the CapeFlyer, this could introduce Massachusetts (including Western Massachusetts) to the idea of taking the train to Montreal — before needing to upgrade the track north of St. Albans or working out the philosophical questions of Higher Speed Rail in Northern New England. As we saw with the Lake Shore Limited versus the Capitol Limited, the longer run time may actually help in this case.

(In the long run, Boston-Montreal service should run via the Vermonter‘s corridor. I’m putting forward the Adirondack routing here because it leverages fully extant services, but realistically any Boston-Montreal overnight service should also run via the Vermonter‘s corridor.)

Burlington, VT

The Adirondack currently stops at Port Kent during the summer. Mere feet from the station is the dock for the ferry to Burlington, VT, which itself lands right in downtown Burlington, a 10-minute walk from the Church St Marketplace, a number of hotels and of course Burlington’s waterfront. The ferry crossing itself is scenic and takes 1 hour. A 7pm departure from Boston would arrive at Port Kent a bit too early to be appealing, but a 9pm departure would arrive at about 6:15am, putting our travelers in Burlington by 7:30am. While it would push arrival in Montreal to 9am, it would add another summer destination to the route.


By splicing the Lake Shore Limited and Maple Leaf schedules, a journey to (downtown) Buffalo would take about 10.5 hours, using current travel times. Now, Buffalo may seem a bit far-flung, but we’ve seen that Buffalo-Chicago can work, and Boston’s CSA at 8.2 million is not that much smaller than Chicago’s 9.8 million, and Boston is a little bit closer. Boston also has the advantage of having Rochester (another 1.1 million CSA on top of Buffalo’s 1.2 million) being that much closer than Chicago.

A 9pm departure from South Station would reach

  • Springfield at 11:30PM
  • Albany at 2:10AM
  • Syracuse at 5:00AM
  • Rochester at 6:15AM

before arriving in Buffalo at 7:30AM the next morning. Again, no track upgrades, and assumes all current stops.

Now, in order to reach Buffalo (to say nothing of similarly large Rochester and only somewhat smaller Syracuse) at a reasonable hour of the morning (not too early), it does mean that Western Massachusetts would be boarding pretty late. If the LSL model holds, Worcester and Springfield would both see passengers, but Pittsfield would be less likely. Likewise, Amsterdam, Utica and Rome all would suffer the “Cleveland” problem of trains stopping in the dead of night. But, Syracuse + Rochester + Buffalo are where the population is, and a 9pm departure could work for those stops.

Niagara Falls

Especially in the summer, certain Buffalo trains could be extended to Niagara Falls, a little less than 1 hour north. A 9pm Boston departure could have you in Niagara Falls by 8:30am the next morning.


Toronto is tricky. Beyond the trackage issues discussed upthread, as well as the need for a customs clearance area at Toronto Union Station, the fact is that the route runs through some 80 miles of Canada, half of which are reasonably dense suburbs of Toronto. A pre-clearance approach would basically require overnight trains to express through all of these suburbs; while most Americans probably aren’t going to Mississauga, an express could cut down on the number of Canadians using the service to get to Boston.

It’s also hard to estimate a hypothetical travel time from the border to Toronto. I guesstimated about 1h20m, assuming something like an average of 50 mph (running non-stop on upgraded track) but that’s a pretty wild guess.

However, all in all, the journey might be something like 13 hours, which should fall into our sweet spot.

A 7pm departure from South Station would reach

  • Springfield at 9:30PM
  • Albany at midnight
  • Rochester at 4:15AM
  • Buffalo at 5:30AM
  • Niagara Falls at 6:45AM

before arriving in Toronto at 8AM the next morning…. maybe.

Note that in order to reach Toronto by the start of the business day, you may have to sacrifice Rochester and certainly Syracuse getting service at a reasonable hour. So, that probably calls for divorcing “local” Buffalo service (which likely would be oriented toward bringing passengers from Buffalo to New England, and so could potentially afford a later arrival in Buffalo on the way “home”) from “international” service to Toronto (which would likely be oriented in both directions) — have the Toronto train leave Boston at 7pm, with the Buffalo train following at 9pm.

One interesting feature of this particular scheduling arrangement is that a 7pm train from Boston could meet a 9:15pm train from New York in Albany at midnight, and either combine the cars (more convenient, but costlier in time) or have a cross platform transfer. In fact, all of the proposals discussed so far could support a meet in Albany. The question would be whether the timing is also convenient for New Yorkers (since NYC-ALB is much shorter than BOS-ALB — read: later departures from NYC).


Cleveland’s CSA is estimated at 3.6 million, which makes it a little bit smaller than Greater Montreal’s 4 million. Based on the Lake Shore Limited‘s schedule (which includes a 1 hour layover in Albany that hopefully could be trimmed by avoiding combining trainsets), a train journey from Boston to Cleveland would take 14.5 hours. Let’s assume we can iron out the layover and call it 14 hours even. This puts it a bit smaller than and a bit further away than Montreal.

A 6:30pm departure from South Station would reach Cleveland the next morning at 8:30am, having hit Buffalo around about 5am. In all likelihood, the reverse journey would probably be the prioritized one — leaving Cleveland at 6:30pm, picking up passengers in Buffalo at about 9:30, and arriving in Boston at 8:30am in time for the work day. If the train were to split in Albany, then Boston arrival would be more like 9am, and New York arrival would be around 7:30.

On its own, a Cleveland-Boston overnighter might be a bridge too far — I’m not sure. However, in the context of the Chicago discussion below and the Buffalo discussion above, it may have its place.


The Lake Shore Limited‘s current schedule is rough for Boston. From Albany and west, the journey is somewhat feasible as an overnighter. The eastbound journey suffers from the time change, but overall it’s still a reasonably good link between Chicagoland and Upstate New York.

But for New England, and for Ohio, the current schedule is not very useful. Too far from New England, and too ungodly an hour for Ohio.

So I propose a second overnight train be added, and be paired with “short-turn” service to Buffalo (and maybe Toronto).

A 6 pm departure from South Station would reach:

  • Springfield at 8:30PM
  • Albany at 11:15PM
  • Buffalo at 4:30AM
  • Erie at 6:00AM
  • Cleveland at 7:45AM
  • Toledo at 10:15AM
  • Elkhart at 12:30PM

And arrive at Chicago at 3:00PM ET/2:00PM CT.

Upstate New York gets stuck with night service, but that can be covered by a Buffalo train that leaves Boston later so as to reach Syracuse during early morning hours. Meanwhile, Ohio can receive passengers during daylight hours, and a reverse train leaving Chicago mid-day can pick up Cleveland passengers around dinnertime, and have them in Boston first thing that next morning.


A comparatively brief note here: in a future where the Inland Route has been rehabilitated, Portland service that reverses out of North Station could be ideal for overnight consideration. The trip from Portland is about 2.5 hours — closer to 3.5 if coming all the way from Brunswick. When added to the aforementioned 9 hour journey to Washington, then you have a ~12 hour journey, and along a corridor that is slightly less saturated (i.e. the north-of-Boston segment). Add an 8pm departure from Portland, arrive in Boston at 10:30, depart North Station around 11pm, arrive in Philadelphia around 6am, and Washington around 8am.

A Review of Proposed B&A Corridor Services

StationBOS-CLE-CHI (14h & 22h)BOS-TOR (13h)BOS-MTL (12h)BOS-BUF (10.5h)
Boston South Station6:00PM7:00PM7:30PM9:00PM
Toronto8:00AM ?
Chicago3:00PM ET

In Conclusion

First, bless you for reading this far, I got totally carried away with this.

Second, I see two major takeaways here.

First — almost all of the above is potentially feasible with existing infrastructure (though possibly not existing rolling stock). Yes, Toronto is a problem. But even before Montreal’s pre-clearance facility opens, that extra hour is not killer on an overnight train. Overall, these proposals are organizational ones that wouldn’t require a drop of concrete to be spilled. The biggest challenge (and it would be a big one, to be sure) would be negotiating with the freight companies that own these tracks. But that is a political problem that ultimately could be solved.

Second — particularly with respect to the BOS-MTL routing, overnight trains potentially give us a “loophole” to get around the ludicrously slow track speeds on many of these corridors. A 12-hour ride to Montreal during the day will never fly except with the most die-hard enthusiasts. But a comfortable ride that you board after dinner in Boston, and awake in Canada? That could be sellable.

Which brings us to the final point: is this sellable? Assuming that the $400 BOS-CHI ticket is strictly a time factor, a BOS-MTL ticket would cost $200-ish. That is probably competitive with a flight to + hotel in Montreal, but it would be tricky. Amtrak would need to price carefully.

Likewise, while there may be some demand for Coach seats on these very long journeys, is there enough Coach demand for a 10 hour overnight supercommute from Buffalo? That’s a tough call.

I leave you with this: between the growing public skepticism in air travel and Amtrak’s renewed push for profitability at the expense of their long-haul routes, there is ripe opportunity here for reimagining the Northeast rail network.

Instead of simply becoming a satellite of New York City, Boston could forge an identity as a place that you can do a convenient “supercommute” to a couple of times a month from Albany, Rochester, Buffalo or Cleveland – or from which you can do a day trip (bookended by Sleeper travel) to Montreal, Niagara Falls, or Toronto — and could give us a distinct identity in the region.

If anyone is interested in the spreadsheets I put together to figure out these schedules, feel free to ask!