Three days in, and this Vuelta still hasn’t shown its real hand. Monaco was a handling puzzle. Stage 2 came down to a bunch sprint. Font Romeu (today’s stage, still unfolding as we write this) is the first summit finish, but it’s one stacked climb into one finish line, a single decision point dressed up as two gradients. Stage 4 is the first day this race stops being one puzzle and becomes four. Andorra la Vella to Andorra la Vella, 104.8km, and somehow 3,024m of climbing packed into a loop that never leaves the principality.

We’ve covered a stacked finish and a single hour-long grind already this year. This is neither. It’s Port d’Envalira, Collada de Beixalís, Coll d’Ordino, and the Alto de La Comella: four categorized climbs, each with a different job to do, separated by descents that give you something back but never all of it. Pace this stage like one long effort and you’ll misjudge which climb is actually dangerous. Pace it like four unrelated climbs and you’ll spend too much on the wrong one. The real skill is treating it as what it is: a full-day energy budget, not a single reserve-for-the-end decision.

Quick Verdict: Pacing a Four-Climb Stage

Climb	Distance/Gradient	Role	Target Effort
Port d’Envalira (cat. 1)	23.3km at 5.1%, summit 2,409m	The marathon opener	78-85% FTP — patience, not power
Collada de Beixalís (cat. 1)	6.5km at 8.5%, ramps to 16%	The shock climb	90-96% FTP — short, steep, honest
Coll d’Ordino (cat. 1)	9.9km at 7%, summit 1,989m	The accumulation test	85-92% FTP — where fatigue actually shows up
Alto de La Comella (cat. 3)	3.9km at 5.9%, summit 1,347m	The launch pad	Full gas if the race is live
Finish	4.8km technical descent to Andorra la Vella	Not a summit finish	Sharp, not spent

Race date: August 25, 2026 Stage distance: 104.8km, Andorra la Vella to Andorra la Vella — 3,024m of elevation gain The lesson: Four different climbs aren’t one long climb with gaps in it. Each one asks a different question, and the finish isn’t even at the top.

Why This Is a Different Problem Than Font Romeu

Font Romeu taught a specific lesson: two climbs stacked with almost no recovery between them are really one long effort in two gradients, and the entire pacing decision is how much you save on the gentle first section for the steeper second one. That’s a two-variable problem. Stage 4 has four variables. And here’s the part that actually changes the math: real descents between three of them. Not full recovery. Nobody free-wheels down a Pyrenean pass at zero watts for twenty minutes and arrives fresh. But a descent isn’t a climb either, and treating this stage like one continuous 100km grind (the mistake that would work fine on a stage with one climb) will leave you making the wrong sacrifice on the wrong ascent.

The shape here is closer to what we wrote about a very different “four climbs” stage back in August: the Tour de France Femmes finale over four ascents of the Col d’Èze. That stage repeated one climb four times, which meant you could calibrate: ride the first three identically, save everything for a fourth ascent you knew in advance would be harder. Stage 4 doesn’t give you that gift. Every climb here is a different road, a different gradient, a different demand, and you don’t get three free reps to learn the pattern before the one that matters. You have to know the shape of the day before you start, because the day won’t teach it to you gently.

The Four Climbs, Broken Down

Port d’Envalira (23.3km at 5.1%, summit 2,409m): This is the highest paved mountain pass in the Pyrenees, and it opens the stage almost immediately out of Andorra la Vella. At just over 5%, it’s not a steep climb by category-1 standards. It’s a long one, closer to 45-50 minutes of steady climbing for most of the peloton. That length is the trap. A gradient this manageable invites riders to sit near threshold for the better part of an hour because it feels sustainable in the moment. It’s the first of four climbs. Ride it like the marathon opener it is: 78-85% FTP, patient, boring on purpose. Whatever you save here you’ll need three more times before the day is over.

Collada de Beixalís (6.5km at 8.5%, ramps to 16%): This is where the character of the stage changes completely. Short, brutally steep, with pitches that hit double digits and spike well past them. There’s no pacing this one the way you paced Envalira. An 8.5% average with 16% ramps doesn’t reward patience, it rewards being able to produce real power while everything above threshold hurts. Target 90-96% FTP here, and expect the RPE to run ahead of the power number on the steepest sections. This is the climb most likely to open the first real gaps of the day, precisely because it punishes anyone who spent too freely on the long, gentle first ascent.

Coll d’Ordino (9.9km at 7%, summit 1,989m): Nearly 10km at a genuinely hard average gradient, arriving after roughly 60km and two descents already in the legs. This is the accumulation test: not the steepest climb of the day, not the longest, but the one where fatigue from the first two efforts actually shows up in the numbers. Riders who paced Envalira and Beixalís correctly should still be producing 85-92% FTP here. Riders who didn’t will find out on this climb, not before it.

Alto de La Comella (3.9km at 5.9%, summit 1,347m) and the descent to the line: Here’s the detail that makes this stage genuinely unusual: the finishing climb is the easiest categorized ascent of the day (a category 3, at km 100 of 104.8) and it isn’t the finish. From the summit, the road drops 4.8km back into Andorra la Vella. That means the stage doesn’t end on a gradient. It ends on a descent, which flips the usual summit-finish logic on its head. Anyone still in contention has every reason to go completely empty on La Comella’s 3.9km, because there’s no climbing left to protect a deficit on afterward — just a technical descent where positioning and nerve matter as much as legs. Ride this climb at whatever you have left, because there’s nothing to save it for.

How Should You Pace a Stage With Multiple Categorized Climbs?

Any event built around several distinct climbs separated by descents (a mountainous gran fondo, a hilly stage race, a long day in the Alps or the Rockies with more than one categorized pass) asks the same four questions, in order:

1. Rank the climbs before you start, not during the ride. Know which one is the longest, which one is the steepest, and which one comes when your legs will actually be tired. Envalira is long, Beixalís is steep, and Ordino is where fatigue actually catches up. Knowing the difference in advance changes how you spend energy on each one.
2. Treat descents as partial recovery, not full recovery. You’ll get some legs back on the way down. You won’t get all of them. Budget the next climb assuming you’re starting from 80% fresh, not 100%.
3. Save your least restrained effort for whichever climb has the least room to fail afterward. On this stage, that’s the last one, not because La Comella is the hardest climb of the day (it isn’t), but because it’s the last place spending everything still pays off before the finish.
4. Don’t assume the summit is the finish. A stage that ends with a descent rewards a different kind of composure than a stage that ends on a gradient. Going empty on the final climb only works if you can still hold a line and read the road on tired legs going back down.

A Workout to Train This

You don’t need four Pyrenean passes to build this pattern. Any ride with three or four hills of different lengths and gradients works, or you can build the shape indoors.

Warm-up: 15 minutes progressive, ending with two 30-second openers at 100% FTP.

Block 1 — “Envalira”: 30-35 minutes at 78-85% FTP (RPE 5-6/10). Long, controlled, deliberately unremarkable.

Descent 1: 5 minutes at 60-65% FTP. Not full recovery — moving, not resting.

Block 2 — “Beixalís”: 8-10 minutes at 90-96% FTP (RPE 8/10), with the final 2 minutes at 100%+ to simulate the steepest ramps.

Descent 2: 5 minutes at 60-65% FTP.

Block 3 — “Ordino”: 15-18 minutes at 85-92% FTP (RPE 7/10), holding steady despite accumulated fatigue.

Descent 3: 5 minutes at 60-65% FTP.

Block 4 — “La Comella”: 6-8 minutes, everything you have left (RPE 9-10/10), no reserve.

Cool-down: 8-10 minutes easy, including a couple of hard accelerations to simulate descending under pressure with tired legs.

Total time: 100-115 minutes.

The test: Compare your power in block 3 (Ordino) to block 1 (Envalira). If it holds close to the same percentage of FTP, you paced the opener correctly. If it drops noticeably, you spent too much on the long, gentle climb and paid for it on the accumulation test. That’s the exact mistake this stage is built to expose. If your FTP number is more than six or eight weeks old, retest it before setting targets off these percentages. A four-climb day is the worst possible place to find out your threshold moved.

What to Watch For Tomorrow

1. Whether the GC gaps open on Beixalís rather than at the finish. With a descent instead of a summit closing the stage, the steepest categorized climb of the day (not the last one) is the most likely place for real separation.
2. How Primož Roglič holds up over four straight climbs. One summit finish is one thing to survive on a compressed build-up. Four categorized ascents in a single afternoon, three days into the race, is a much harder question.
3. Whether Pogačar’s margin from Monaco survived Font Romeu today. We won’t know that until this stage starts, but a rider defending any lead into a day with this much climbing density has more to lose than gain.
4. Who descends well off La Comella. A stage that finishes downhill instead of uphill rewards nerve on tired legs as much as it rewards climbing legs. Watch the final 4.8km as closely as the climb above it.
5. The overall shape of the front group entering Andorra la Vella. If it’s still large after three cat-1 climbs, the full route breakdown suggests the real GC sorting on this route is still further up the road, but a stage this dense in climbing rarely leaves a big group intact.

The Bottom Line

Stage 4 isn’t harder than Font Romeu because any single climb on it is worse. It’s harder because it asks four separate pacing questions instead of one, with descents in between that give back just enough energy to tempt you into spending it wrong. Ride Envalira like the long, patient opener it is. Respect Beixalís for what its gradient actually does, not what its length suggests. Recognize Ordino as the climb where earlier mistakes get exposed. And go empty on La Comella, because for the first time all week, there’s no uphill finish left to protect.

For anyone riding a mountainous event with more than one climb on the profile, the transfer is direct: rank your climbs before the ride starts, and budget descents as partial recovery rather than full recovery. Save your least restrained effort for whichever ascent leaves the least room to recover afterward. That’s the whole test Andorra is setting up tomorrow, and it’s the same one waiting on any long day with more than one mountain on the map.

---

Stage distance, elevation gain, and climb data from La Vuelta’s official stage 4 page. Additional route and climb context from IDL ProCycling’s 2026 Vuelta route preview. Training targets are estimates based on power-based training principles. Adjust for your own tested FTP and conditions.