Zwift's New York, and the City It Is Based On

I have ridden Zwift's New York a great many times. I have never ridden the real one.

That split is worth stating at the top, because it decides what this article can honestly be about. I can tell you what the in-game world does to your legs, which routes suit which session, and why urban riding is an interval workout wearing a commuter's clothes. I cannot tell you what it feels like to cross the Brooklyn Bridge at seven in the morning, and I am not going to pretend otherwise.

So this is mostly a training article. The through-line is that stop-start city riding and Zwift's New York world happen to demand the same physiological quality, which is the ability to produce a large number of hard accelerations without the gaps between them turning into recovery. That quality is trainable indoors, and most riders train the opposite of it by accident.

What Zwift's New York actually is

Zwift's New York is two worlds stacked on top of each other, and they ride nothing like one another.

At ground level there is Central Park, laid out with roughly the geometry of the real thing. It is a rolling loop rather than a flat one, with a long drag on the western side and a short sharp rise near the northern end that will take more out of you than its length suggests. Ridden at a steady effort it is one of the better endurance loops in the game, because the gradient changes often enough to keep you honest without ever becoming a climb.

Above the park there is a network of elevated transparent roads that exist only in the game. These carry the steepest gradients in the world and the ramps on and off them are abrupt. The transitions are what make the world interesting to train on: you leave a rolling park loop, hit a short steep ramp, sit up on a fast elevated section, and then drop back down. The pattern repeats.

Route names, distances and elevation figures change with game updates often enough that quoting them here would be out of date within a season. Check the route list in-game for current numbers. What does not change is the shape of the effort each type of route asks for, and that is the part worth planning around.

Three kinds of route, three different sessions

Whatever the current route list looks like, everything in the world falls into one of three buckets.

• Park-only loops. Rolling, continuous, no long climb. These are endurance and tempo routes. Nothing forces a surge, so if you ride one hard it is because you chose to.
• Mixed park and elevated. The park loop with one or more trips up onto the glass. These are the interval routes, whether or not you intended to do intervals, because the ramps are steep enough that you cannot hold a steady power through them without changing effort.
• Elevated-heavy. Repeated short steep sections with fast recovery between them. Closest thing the world has to a structured over-under session with scenery attached.

The mistake I made for a long time was picking the second kind of route and then trying to ride it at a single steady wattage. It does not work. The ramps force you above threshold whether you like it or not, and the fast sections afterwards make it almost impossible to keep the average where you wanted it. You finish having done a poor version of two different sessions.

Why riding in a city is an interval session

Here is the part that is worth doing the arithmetic on, because it explains why people come back from a week of city riding more tired than the distance suggests they should be.

Getting a bike and rider back up to speed costs energy, and that cost is set by kinetic energy, which scales with the square of the speed. Take a rider and bike at 85 kg total, accelerating from a standstill to 25 km/h. 25 km/h is 6.94 m/s, so:

Two thousand joules does not sound like much. The point is how often it repeats. Forty stops in an hour, which is not an unusual number for dense city riding, is roughly 82 kJ spent purely on getting back up to speed. An hour of steady riding at 200 W is about 720 kJ in total, so those accelerations are on the order of a tenth of the whole ride, and every one of them is delivered in a short, hard burst rather than spread out.

That is the real physiological signature of urban riding. The average power looks unremarkable. The distribution is brutal. And it is exactly what Zwift's New York world produces when you ride the ramps, which is why the world is a better preparation for city riding than a flat route would be, despite the flat route matching the terrain more closely.

Why the average power on that ride lies to you

Two rides can finish with the same average power and cost you completely different amounts. This is the single most useful thing to understand about stop-start riding, and it is why a city ride wrecks you while a steady ride of the same average leaves you fine.

Consider two hours, both averaging 180 W.

The anaerobic contribution is the part that matters. Every effort above threshold draws on a capacity that refills slowly, and refilling it requires riding easier than you were before. In a city you rarely get to, because the next junction arrives before you have recovered from the last one. Do that eighty times and you finish genuinely fatigued from a ride whose average says you were barely trying.

This is also why the average power figure is worth ignoring on rides like this. Look at time above threshold instead, or at the number of efforts. Those describe what actually happened.

A session that actually trains it

Repeated hard accelerations with short recoveries is a well-defined training stimulus, and you do not need a bespoke workout file for it. Pick a route with several elevated sections and ride it deliberately.

• Warm-up: 15 minutes easy on the park loop, with three 20-second openers in the last five minutes.
• The work: every time a ramp arrives, get out of the saddle and take it hard, aiming for roughly 150% of FTP for the 20 to 40 seconds it lasts. Do not pace it. The whole point is that you did not get to choose when it started.
• Between ramps: ride at endurance pace, not recovery pace. This is the part people get wrong. If you soft-pedal back to fresh between efforts you are training something else entirely.
• Volume: start at eight to ten ramps. Build toward fifteen or twenty across a block.
• Cool-down: 10 minutes easy.

The session is harder than it reads, mostly because of the third bullet. Holding endurance power between maximal efforts is the specific thing that city riding demands and that structured indoor training usually omits, because a normal interval session gives you genuine rest between reps.

If you want a number to judge it by, watch how much your power in the last three ramps drops relative to the first three. Losing more than about 15% means you started too hard or you took too few of them. That is a judgement I have arrived at from riding these sessions rather than from any dataset, so treat it as a working rule and not a measurement.

What the game cannot prepare you for

The engine transfers. Almost nothing else does, and it is worth being blunt about that, because the gap is where people get hurt.

• Braking and reading traffic. Zwift has no brakes and nothing to avoid. The single most important skill in city riding gets zero practice indoors.
• Starting from a track stand or a foot down. Repeatedly, in cleats, next to a kerb.
• Surface. Expansion joints, tram rails, drain covers and potholes.
• Cognitive load. There is a whole article on this site about how decision quality degrades as effort rises. Indoors that costs you a race position. In traffic it costs more than that.

The honest framing is that a winter on Zwift means you will arrive fit enough for a day in the city and under-practised at everything except the pedalling. Give yourself a couple of quiet rides somewhere forgiving before the traffic gets a vote.

If you are actually going

I am outside my expertise here, so this section is short and consists only of things that are widely agreed on by people who do ride there. Treat it as a starting point for your own research rather than as local knowledge.

• The Hudson River Greenway on Manhattan's west side is largely separated from traffic and is the usual recommendation as a first ride and as a connector between other rides.
• The Central Park loop is one-way for cyclists and is busiest in the middle of the day. Early morning is the consistent advice.
• Protected infrastructure varies enormously block to block. Planning around the corridors that have it, rather than taking the direct line, is what most first-timers say they wish they had done.

Keeping the pattern after the trip

The reason I keep coming back to this world in winter is that the repeated-acceleration stimulus is genuinely useful and genuinely unpleasant, which is a good combination for something you would otherwise talk yourself out of. A structured over-under file does the same physiological job, and I skip it far more often, because a workout screen asking for the eleventh rep is easier to quit than a ramp that is simply there in front of you.

That is not a physiological argument. It is a behavioural one, and it is the honest reason the world earns a place in my week. If you want the physiological case for choosing routes deliberately rather than by mood, it is set out in the article on route selection.

The short version

• Zwift's New York is a rolling park loop plus a set of steep elevated ramps that exist only in-game.
• City riding is physiologically an interval session: around 2,050 J per acceleration from a stop, forty of them an hour, roughly a tenth of the ride delivered in short bursts.
• The ramps reproduce that pattern better than a flat route would, despite matching the real terrain less well.
• Ride endurance pace between ramps, not recovery pace. That is the whole session.
• None of the handling, braking or traffic-reading transfers. Plan for that separately.