First Pages or a Chosen Scene? Prepare the Screenplay Sample the Lab Actually Requests
First Pages or a Chosen Scene? Prepare the Screenplay Sample the Lab Actually Requests
You have a scene at page 63 that works. The first five pages are quiet, careful, and — you suspect — boring. The lab asks for a sample. Is this the moment to lead with the good stuff?
Only if the lab said you could. The instruction is the permission.
Most versions of this problem collapse into one question — what is my best writing? — when the actual question is what did they ask for? Those come apart at exactly the point where writers get into trouble: when the best writing sits somewhere the instructions did not invite you to go. So the order of operations is read first, choose second. Taste does not get a vote on the specification.
Read the requirement literally, and record which cycle you're reading
Application calls tend to arrive in a few recognizable shapes, and they are not variations on a theme.
Some prescribe the opening: the first five pages, the first ten, the first act. Some allow an excerpt you choose, usually inside a page cap. Some want something else specifically defined — one complete scene, a page range from a particular draft, or an alternate artifact such as a treatment. Occasionally a call specifies two different documents for two different tracks, and it matters which one you're applying to.
Write down which shape you have, along with the page limit, the file format, whether the title page counts against the limit, and whether the call treats a screenplay and a treatment as interchangeable. Those details are the specification. Everything after this is compliance, and compliance is not a lesser art — it is the only part of the submission you fully control.
The word sample does not create choice. Neither does your best pages, which appears often and usually means the best pages you can send within the limits we just described. It rarely means skip the opening. And do not inherit a rule from somewhere else. A program that asked for a chosen scene last cycle, or a friend's program that wanted twelve pages, tells you nothing about this call. Requirements change between cycles, and the call is the only authority in the room.
If a sentence is genuinely ambiguous in a way that changes what you send — a page limit that may or may not include the title page, a rule that could mean the opening or any episode — ask the program. Programs answer this constantly. A three-day wait is cheaper than a rejected file.
When it asks for the opening, prepare the opening
Some calls say it plainly. Sundance Institute's 2027 Development Track, in instructions recorded for that cycle in September 2025, asked for the first five pages of the screenplay and instructed applicants not to exceed that opening excerpt. The call's published closing date was May 12, 2026. The treatment route belongs to the Sloan Commissioning Grant, not to the Development Track.
That is one dated cycle, not a rule about labs, and not an invitation that is currently open. Check the live call before you submit anything. But it is a clean example of the first shape, and two useful things follow from it.
The first is what it asks for: pages 1 through 5 of the current script, exported as pages 1 through 5. Not the five best pages. Not pages 1, 2, 3 and then a jump. Not a later scene with a note explaining that the opening isn't representative yet. If the opening is five pages, it ends wherever page five ends — possibly mid-scene, possibly mid-line. An opening that is still assembling itself at page five is normal, not a defect you need to conceal.
The second is what to do when the opening doesn't work. If pages 1–5 are genuinely weak, you have a screenplay problem. The fix is revision, or a deliberate decision to wait for a cycle whose requirements fit the draft you actually have. Substituting different pages isn't presentation; it's a different document from the one requested, handed to readers who will notice. It also costs you the thing a sample is uniquely good at showing — that you can follow an instruction, and that your script stands on its own from the first page.
Whether a title page consumes one of the five is settled in some calls and genuinely unclear in others. With a five-page limit, the difference between four pages of script and five is not a technicality. Ask.
When choice is expressly allowed, choose a complete dramatic movement
Now the other case. The call says some version of: send up to twelve pages from anywhere in the screenplay. Those words are the only thing that permits a later scene. Under them, your job changes from extraction to construction.
You are looking for continuous pages that contain a movement. Something begins, pressure accumulates, and by the last page the situation is different from the one on the first. Setup and consequence both inside the excerpt — not setup here and consequence somewhere the reader can't see.
Where the passage starts and ends is the entire craft decision. A speech that dazzles in isolation usually depends on the scene that produced it: the relationship, the debt, the thing promised three scenes earlier. Lift it out and the reader gets the temperature without the weather. The better choice is often to back up until the passage can stand alone, which sometimes means accepting that your most quotable page isn't in the excerpt at all.
The excerpt does not have to explain the feature. It has to do dramatic work. A paragraph of prose context stapled to the front is not part of a screenplay and is not permitted in most calls. If the format allows a short note, write it in the terms the call allows. If it doesn't, trust the pages. The synopsis is a separate artifact with a separate job, and it cannot carry the excerpt's drama for it.
Check the file you actually extracted
This is the unglamorous part, and it is where compliant selections quietly go wrong.
Read the first and last pages of the PDF you're about to send. Then open the source script and read the page before your first page and the page after your last, to confirm the boundaries fall where you think they do. Count the pages under the call's rule, applying whatever you decided about the title page. Confirm the file came from the draft your synopsis describes — a newer script paired with an older synopsis is a common, entirely avoidable contradiction, and readers read both.
Then look for strays. Comments. Tracked changes that exported as text. A page in revision color. A scene you replaced months ago, still living in a file you forgot was open. A collaborator's note in the margin. Exports pick up what writers forget.
Keep the decision where it belongs, too. Choosing pages that answer the instruction is your job. It is not a forecast. You can send a scrupulously compliant five pages into a competitive pool and hear nothing back. What compliance does is remove one reason to be rejected — no more, and no less.
The same script, two specifications
Let me put concrete pages on this. The screenplay below is invented for the exercise — no production, no writer, no file — and so is the second specification. Only the Sundance instruction is a real, dated observation.
Low Water, a 104-page feature. Marisol, forty-one, takes over her late father's municipal swimming pool in a town under water restrictions. The council has voted to close the pool at the end of summer. Her brother Elias sits on that council. The opening is restrained: a woman unlocking a gate before dawn, a laminated notice taped inside the glass, a pump room, a councilman arriving early with decommissioning paperwork. On page five, Marisol gets into the water fully clothed and goes under.
The sequence everyone likes runs pages 59 to 72. It opens quietly — two pages of Marisol letting neighborhood kids swim after dark — then page 61 has her alone at the meter with a call to the water department, and page 62 brings Elias to the gate at eleven at night with a photograph of that meter. The argument that follows is about their father, the will, and his vote. It ends with Elias taking the red zip tie off the intake valve and walking out, leaving the gate open. Fourteen pages, start to finish.
Specification A: the first five pages. Pages 1–5, as they stand. Quiet, mostly visual, one page of dialogue. The dramatic question is on the page — the pool is closing. The reader meets someone who does one slightly strange thing in the last two lines. That is a legible opening, and if the writer believes it isn't, the diagnosis is a screenplay problem: the opening needs work, or the story starts in the wrong place. It is not an excerpt problem. An instruction like Sundance's, on its recorded terms, is answered by that file and by nothing else.
Specification B: an invented free-choice rule. For the sake of comparison, suppose a program that does not exist — call it the Lakebed Lab — publishes: Submit up to twelve continuous pages from anywhere in your screenplay. This is a teaching construction. No real program's rule is being described.
Twelve pages against a fourteen-page sequence forces a boundary decision. Two boundaries present themselves, and only one of them works.
Pages 61–72. Page 61 establishes the meter and the call. Page 62 brings Elias. The argument runs, the information lands, the decision happens, and the last image is the decision itself. Twelve continuous pages with setup and consequence both inside. This is the extract that answers the invented rule.
Pages 61–71. One page shorter, comfortably inside the limit, and wrong. It ends on the moment before Elias decides — he's holding the zip tie, saying he could lose his seat for this. His answer is on page 72. Stop there and the reader has watched a man refuse. Same characters, same dialogue, opposite story. Fitting the limit is not the same as earning the space. The eleven-page version is more comfortable and less true.
The version that satisfies neither. Pages 1–3, then 61–63: six pages in two discontinuous chunks. Against Specification A it exceeds five pages, and the second chunk isn't the opening. Against Specification B it isn't continuous. It's a highlight reel with the connective tissue removed — precisely what the first instruction forbids and what the second one doesn't invite either. If you find yourself drafting a parenthetical to bridge the gap, that's the signal. You've built the wrong artifact.
Notice what happened there. The same writer, the same draft, made two different correct choices and one incorrect one. Nothing about the screenwriting changed. The instruction did.
What you should be holding when you press send
A file, and the sentence it answers.
Say the instruction to yourself before you export: first five pages, or up to twelve continuous pages from anywhere. Then check that the file matches that sentence — source draft, boundaries, page count, no stray material, synopsis describing the same version. If it matches, the page selection is finished, and whatever you think of page one is a separate conversation you can have with your own draft afterward.
If it doesn't match, you have three honest options: revise the script so the requested pages are pages you're glad to send, ask the program about the specific ambiguity, or wait for a cycle whose instruction fits the work you have. None of those is send something else and explain why. That fourth option doesn't exist, no matter how good page 63 is.
Frequently asked questions
Does an invitation to send 'your best pages' allow me to skip a requested opening?
No. The body says 'sample' does not create choice, and 'your best pages' usually means the best pages you can send within the described limits. It rarely means skip the opening; the instruction is the permission.
What details should I record from an application call before selecting pages?
Record which shape the call uses, the page limit, file format, whether the title page counts against the limit, and whether a screenplay and treatment are treated as interchangeable. Those details are the specification.
If the call is genuinely ambiguous in a way that changes what I send, what does the article recommend?
Ask the program. Programs answer this constantly, and a three-day wait is cheaper than a rejected file.
What should I do if my first five pages are weak but the call asks for the opening?
Treat it as a screenplay problem. Revise the script or wait for a cycle whose requirements fit the draft. Substituting other pages is a different document from the one requested and costs the chance to show you can follow an instruction.
When an excerpt from anywhere is expressly allowed, what makes a strong selection?
Choose continuous pages that contain a dramatic movement: something begins, pressure accumulates, and the situation is different by the last page. Setup and consequence should both be inside the excerpt; where it starts and ends is the craft decision.