We’ve been pointing at this stage since before the race even started. Our amateur training guide called Stage 20 the hardest day of the route back in August, and every mountain post since has repeated some version of the same line: whatever happens on the stages around it, this is the one that actually decides the Vuelta. Stage 20 of the 2026 Vuelta a España is 186.8km from La Calahorra to an unprecedented HC summit finish at Collado del Alguacil, with roughly 4,849m of climbing packed into five categorized ascents along the way. That’s the queen stage. Tomorrow’s Madrid finale is a procession. This is the race.

Enric Mas came into this final week leading Félix Gall by 2:06 and Primož Roglič by 2:10, a gap built almost entirely in the mountains, according to ProCyclingUK’s state-of-the-race analysis. He’s since defended that lead through the one stage built specifically to attack it, Stage 18’s flat time trial, and through 210.8km of accumulated fatigue up Peñas Blancas. Whatever the exact number looks like after those two tests, one thing hasn’t moved: Stage 20 is where this Vuelta gets decided, not tomorrow’s ceremonial roll into the Alhambra.

Quick Verdict: Pacing the Vuelta’s Queen Stage

Section	Distance/Gradient	Role	Target Effort
Puerto de Los Blancares (cat. 3)	7.7km at 3.7%, summit km 50.7	Warm-up climb, not worth spending on	65-72% FTP
Puerto de El Purche, 1st ascent (cat. 1)	8.2km at 8.1%, summit km 94.3	First real test, brutal opening kilometers	78-85% FTP
Puerto de El Purche, 2nd ascent (cat. 1)	8.2km at 8.1%, summit km 125.5	Same climb, legs that have already done it once	80-87% FTP, watch the fade
Puerto de El Duque (cat. 1)	7.4km at 8.3%, summit km 155.4	Narrow, steep, includes a 20%-plus ramp (save something)	82-90% FTP, RPE 7
Collado del Alguacil (HC, the finish)	8.3km at 9.8%, summit km 186.8	Double-digit gradient almost the entire way, after 4,849m of prior climbing	88-95%+ FTP, RPE 8-9

Race date: September 12, 2026 Stage distance: 186.8km, La Calahorra to Collado del Alguacil, roughly 4,849m of elevation gain, the hardest day of this Vuelta The lesson: Every climb before the finish is a tax, not a test. Pay it as cheaply as possible, because the only climb that matters arrives after all four others, at 9.8% average, with nothing left to hide behind.

The Stage the Whole Route Has Been Pointing At

Grand Tour organizers love calling stages the hardest of the race. Most years that’s marketing. This one earns it on the numbers alone: five categorized climbs, a double ascent of a category 1, and a summit finish that’s never hosted a Vuelta stage before, according to Domestique Cycling’s stage-by-stage guide. Collado del Alguacil going straight in as an HC finish, with no prior Vuelta history to calibrate expectations against, is exactly the kind of unknown that makes GC teams nervous heading into a stage they can’t fully prepare for.

The context matters as much as the profile. Stage 14’s Sierra de la Pandera taught this race that a short, steep wall can blow a GC group apart in two kilometers. Stage 19’s Peñas Blancas taught it that a long, honest grind on tired legs does something different. It doesn’t ambush anyone, it just grinds down whoever’s weakest at holding a number for 45 minutes. Collado del Alguacil doesn’t pick one of those problems. It’s long enough to matter and steep enough to punish, arriving on the single biggest climbing day of the entire route. If Mas is going to lose the red jersey in 2026, the terrain says it happens here.

The Route, Broken Down

Puerto de Los Blancares, 7.7km at 3.7% (cat. 3, summit km 50.7): A gentle opener, topping out at 1,298m. Not hard enough to justify racing it, but early enough that a team riding it aggressively can start the day’s fatigue account before the stage has really begun.

Puerto de El Purche, 8.2km at 8.1% (cat. 1, summit km 94.3 and again at km 125.5): The stage’s first real fight, and riders do it twice. El Purche’s opening kilometers are the brutal part: a run of 9-11% pitches, with one stretch reaching 17%, according to cyclingstage.com’s route profile. Climbing that twice, 31km apart, is less about the second ascent being harder on paper and more about it landing on legs that already paid for the first one.

Puerto de El Duque, 7.4km at 8.3% (cat. 1, summit km 155.4): New to this race entirely, a narrow forest road with a steep opening section and more than two kilometers averaging over 10%, including a ramp past 20%. It’s the kind of climb that rewards positioning as much as raw power, because there’s little room to recover a bad move on a road this tight.

Collado del Alguacil, 8.3km at 9.8% (cat. HC, the finish): The decisive climb, and the numbers alone explain why. A dead-end road into the Sierra Nevada, holding double-digit gradient for almost its entire length aside from the opening kilometer, per cyclingstage.com’s profile. Unlike Peñas Blancas’ rideable lower two-thirds, there’s no honest section to settle into here. The climb asks for a sustained, near-threshold effort from close to the base, on legs that have already climbed four other ascents and roughly 4,033m before the final ramp even starts.

Why a Wall After 4,849 Meters Is a Different Problem

Two stages have already taught this Vuelta’s hardest lessons about climbing. Pandera showed what a short, violent ramp does to a fresh-ish peloton: an anaerobic, standing effort where torque matters more than sustained threshold power. Peñas Blancas showed what a long, honest grind does on legs already carrying eighteen days of racing, a discipline problem rather than a power problem, solved by starting conservative and holding a number.

Collado del Alguacil doesn’t let you pick which lesson applies. At 8.3km, it’s long enough that you can’t treat it like a two-minute wall — you need a real threshold-level plan for the climb, not a sprint. But at 9.8% average with double-digit gradient almost throughout, it’s steep enough that “just hold your number” isn’t the whole answer either, because the road itself keeps trying to push that number up whether you want it to or not. And it arrives after 4,849m of prior elevation gain, more total climbing in one stage than this Vuelta has asked for on any other day, including Peñas Blancas’ marathon summit finish. A rider who paced the four lead-in climbs like they didn’t matter still has to solve a genuinely hard climb at the end. A rider who didn’t will find out exactly how unforgiving 9.8% is on legs with nothing left to give.

How Do You Pace the Vuelta’s Hardest Summit Finish?

A stage built around five climbs and a double-digit HC finish punishes a specific mistake: treating any of the first four ascents as worth racing, when the only climb that actually decides the Vuelta is the fifth one.

1. Ride the first four climbs at the minimum effort the group allows. Puerto de Los Blancares, both ascents of El Purche, and Puerto de El Duque combine for over 31km of categorized climbing before the finish even starts. None of them are worth spending real matches on.
2. Respect the second ascent of El Purche more than the first. Same gradient, same distance, but a completely different physiological cost, because it lands on legs that already climbed it once, 31km earlier. Riders who ride the first ascent hard often don’t notice the bill until the second one.
3. Start Collado del Alguacil below the number you think you can hold. There’s no honest lower section here to ease into, unlike Peñas Blancas. The climb is steep almost immediately, which makes an over-eager opening kilometer more expensive than it would be on a gentler ascent.
4. Expect your real threshold to be lower than your tested one. After four prior climbs and 186km of racing, the power you can sustain for the final 8.3km is not the number from a fresh FTP test. Pace off perceived effort: RPE 8 should feel like RPE 8, not like a number you’re forcing.
5. Save nothing for tomorrow. Unlike every stage before it in this final week, there’s no “protect something for the next hard day” calculation here. The Madrid stage is processional. Whatever’s left in the legs after Collado del Alguacil is meant to be spent today.

A Workout to Train This

You don’t need five real mountains to train this shape. A long ride with several unremarkable climbs, then one genuinely hard sustained effort at the end, teaches the same lesson: spend nothing early, because the finish doesn’t forgive a rider who arrives with an empty tank.

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

Block 1 (“Blancares”): 12-15 minutes at 65-72% FTP (RPE 4/10), easy and controlled.

Block 2 (“El Purche, 1st time”): 15-18 minutes at 78-85% FTP (RPE 5-6/10). Real effort, but not a race.

Block 3 (“El Purche, 2nd time”): 15-18 minutes at 80-87% FTP (RPE 6/10), the same target as Block 2, but notice whether it costs more to hold. If it does, that’s the point.

Block 4 (“El Duque”): 12-14 minutes at 82-90% FTP (RPE 7/10), with one unscheduled 45-second surge to 105% FTP to simulate a steep, narrow pitch where you can’t pick your own line.

Block 5 (“Collado del Alguacil”): 25-30 minutes at 88-95%+ FTP (RPE 8-9/10), held as evenly as you can manage from the first minute, no easing in, because the real climb doesn’t offer one either.

Cool-down: 10 minutes easy.

Total time: 100-120 minutes.

The test: Compare your power in the first five minutes of Block 5 against the last five. A climb this steep punishes anyone who starts too hot far more than a gentler ascent would. If you’re fading hard by the final five minutes, you likely went out too fast, the same mistake that turns a manageable HC finish into a blown-up one. If your FTP number is more than six or eight weeks old, retest before trusting these targets. A double-digit average gradient is a bad place to discover your number was wrong.

What to Watch For Today

1. Whether anyone attacks on the first ascent of El Purche. It’s early and it’s tempting, but a long-range move with four climbs still to come is a gamble against a field that hasn’t shown its full hand yet.
2. How the group looks on the second ascent of El Purche. Same climb, same gradient, and the riders who visibly struggle more the second time around are showing exactly who’s paying for the accumulated fatigue of this final week.
3. Whether Roglič or Gall need to distance Mas before Collado del Alguacil. A double-digit HC finish gives a strong climber real terrain to build a gap on, but it also means Mas doesn’t need to match every acceleration, just limit the damage the way he did in Stage 18’s time trial.
4. Who’s still riding their own number in the final two kilometers. At 9.8% average, the riders who paced the first four climbs conservatively should have something left when the gradient stops forgiving anyone.

The Bottom Line

Every stage in this final week has been building toward this one, and the terrain backs up the hype for once. 186.8km, 4,849m of climbing, five categorized ascents, a double passage of El Purche, and an HC summit finish that’s never decided a Vuelta before because it’s never hosted one. If Enric Mas is still in red after today, it’s because he rode the first four climbs like they didn’t matter and the fifth one like it was the only thing that did. If he isn’t, this is where it happened — not on a time trial built to catch him, not on a long grind built to wear him down, but on a wall that asks for both sustained power and raw steepness at the same time, after more climbing than this race has demanded on any other single day.

The transfer to your own training doesn’t need a Grand Tour behind it. Any ride with several moderate climbs early and one genuinely hard, steep effort at the end tests the same skill this stage does: spend nothing you don’t have to on the climbs that don’t matter, so there’s something left for the one that does.

---

Stage distance, climb data, and elevation figures from La Vuelta’s official stage 20 page. Stage significance and route context from Domestique Cycling’s Vuelta a España 2026 stage-by-stage guide. Climb gradient detail and profile notes from cyclingstage.com’s stage 20 route page. GC standings entering the final week from ProCyclingUK’s state-of-the-race report. Training targets are estimates based on power-based training principles. Adjust for your own tested FTP and conditions.