The real dates (verify them, do not trust them)
Here are the announced 2026-27 US administrations. Treat this list as a starting point, not gospel: dates and registration deadlines shift, and the College Board is the only current source. Confirm each one on the live College Board calendar before you commit a dollar or a plan to it.
- August 22, 2026
- September 12, 2026
- October 3, 2026
- November 7, 2026
- December 5, 2026
- March 6, 2027
- May 1, 2027
- June 5, 2027
Two things that list hides. First, the registration deadline for each date lands roughly a month before the test, with a later, more expensive window after that. Do not memorize a deadline you read here: check the current one for your date, because that regular deadline is the day your options quietly close. Second, none of these dates is the deadline that actually governs your score. That one is earlier, and you set it yourself.
The 4-week cycle (why the registration deadline is a decoy)
A test date only produces a better score if a full prep cycle happens before it. The cycle has a fixed minimum length, and it looks like this:
- Week 1, diagnose. Take one full practice test. Sort every miss into buckets by question type, the way the plateau playbook lays out: punctuation, transitions, words-in-context, evidence, inference. Most students expect fifteen scattered problems and find two or three fat buckets instead.
- Weeks 2 and 3, drill. One bucket per short daily session, roughly twenty minutes. After every miss, write the one-sentence real reason (the One-Sentence Autopsy; "careless" is banned). This is where points are actually made, and it is the slowest part, so it sets the floor on the whole timeline.
- Week 4, simulate. Now the full timed tests come back, to integrate what you fixed and build stamina, not to re-measure what you had not touched yet.
That is about four weeks, and it does not compress well: the drilling weeks depend on spacing between sessions, and spacing is time you cannot buy back. So the deadline that decides your score is four weeks before test day. Registering on time keeps a seat available. Starting the cycle on time is what fills it.
Run this self-test today. Take any date you are eyeing and count back four weeks on a calendar. Is that landing date comfortably in the future, with slack for a bad week? If yes, that date is reachable. If the four-weeks-back mark is already behind you or lands tomorrow, that date is too soon for a real cycle: either you start diagnosing today, or you pick the next date and do it properly.
Junior-year sequencing (the clean order)
The cleanest sequence for most students is a first real sitting in the spring of junior year (March, May, or June) with the fall of senior year held in reserve. That structure does two things. It gets a genuine, cycle-backed score on the board while there is still runway, and it leaves a retake window open before senior-fall application deadlines without forcing a panic sitting.
What breaks this is treating the spring date as a casual trial. If you walk in without the four-week cycle behind you, you have not bought information, you have bought noise. Build the actual plan around the date you choose: a six-week R&W plan if you have the runway, a compressed two-week plan if the date is already close and you have decided to commit to it anyway.
Retake logic
Before you book a retake to chase a few more points, learn the fact that governs the decision: College Board reports section scores as a range rather than a single point, because scores vary between administrations. Two sittings of the same student with the same underlying ability routinely land apart on the instrument alone.
That gives you a hard rule for retakes: a retake that moves you a small amount is noise rather than improvement. A 1290 that becomes 1310 is the same score wearing a different outfit. You did not get better; the measurement wobbled. So retake only when both of these are true: you have run a fresh diagnose-drill-simulate cycle since the last sitting (new inputs, not just a new Saturday), and the gap you are chasing is bigger than the noise floor the instrument can even see. If your last score is already inside striking distance of your target, the honest move is often to stop, not to keep feeding the machine. The plateau playbook covers why re-measuring is not the same as improving.
The one-date-as-last mindset
Here is the quiet leak in date-picking. The moment you tell yourself "I'll wing this one and get serious for the next date," you have already lowered your effort on the date in front of you. A backup date does not add a safety net; it drains the seriousness out of the attempt you are actually about to make. The option to retake becomes a reason to under-prepare.
So flip it. Treat every date you sit as if it is your last one. Prep once, with a full cycle behind you, and walk in expecting this to be the number that counts. This is not about pressure. It is about not paying a registration fee to gather noise. If a genuine retake is warranted afterward under the retake logic above, fine, but that is a decision you make from a real result, not a hedge you plan from the start.
Start the cycle
The mistake is agonizing over October versus November. The decision that matters is not which square you circle; it is whether a real four-week cycle sits behind it. The date is downstream of the diagnosis. So do the thing that unblocks the choice: find out what your misses are actually made of, then let the calendar fall out of the answer.
Start the cycle now: the free simplesat diagnostic sorts your misses into buckets in about ten minutes, no signup, so you can pick a date a real four-week cycle can actually reach.