How to Win the ISEF Interview: The 25 Points Nobody Practices

Here is the most lopsided fact in science fair preparation: your poster is worth 10 points and the interview is worth 25, and students routinely spend twenty hours on the board and zero rehearsing the conversation. At ISEF you are interviewed at least four times, 15 minutes each, by different judges, and the same interview weighting flows down through the regional and state fairs that feed it. This is how to prepare for the part of the score that is actually the largest.

What judges are listening for

This is not a mystery. The published criteria tell you what the interview assesses:

• Clear responses to questions
• Understanding of the basic science underlying the project
• Interpretation and limitations of the results
• Degree of independence in conducting the work
• Recognition of potential impact in the field
• Quality of ideas for further research

Notice what is not on that list: charisma, polish, confidence theater. The interview is a technical conversation scored on understanding. That is very good news for prepared students and very bad news for coached-sounding ones.

The four questions that decide it

Across regional fairs, states, and ISEF itself, four questions do most of the sorting. Prepare real answers to these and you are ahead of the majority of the room.

1. "What is the biggest weakness in your study?"

Never say "none." Every judge knows there is one. The question tests whether you do.

Weak: "I think the study was pretty thorough, maybe more data would help." Strong: "The sample in my test condition is small enough that I don't claim an ordering within my top results, only the separation from baseline, which is larger than my method's error. Here is the experiment that would resolve it."

Naming your limitation with precision is not a confession. It is the single highest-scoring move available in the interview, because limitations are literally on the scoring criteria.

2. "Why this method and not another?"

This question exposes outsourced projects instantly. If your mentor chose the method and you cannot reconstruct why, the judge finds the edge of your understanding in one step.

Weak: "It's the standard approach for this kind of problem." Strong: "Two methods fit. I used X because my data has [property], which breaks Y's core assumption. The tradeoff is [cost], which shows up in my results as [effect], and here is how I handled it."

3. "What would you do differently?"

Weak: cosmetic answers. "Start earlier, get more data." Strong: structural answers. "I would design the validation split differently from the start, because I discovered late that my original split could leak information between training and testing. I reran it correctly, and both sets of numbers are on the board."

Judges love this answer because it is what real researchers sound like.

4. "What is the next experiment?"

Vague futures cost more here than students expect, because "quality of ideas for further research" is an explicit criterion.

Weak: "Expand the study and explore more directions." Strong: One named experiment with a design: "An in vitro assay measuring [specific thing] against [specific control]. If it shows [X], that confirms my mechanism. If not, my alternative explanation is [Y]."

The independence trap, especially for mentored projects

Judges explicitly score how much of the project is yours. If you worked with a mentor, a lab, or a program, that is not a problem, unmentored projects are rare at the top level. The trap is a student who cannot explain a decision that someone else made.

The exposing questions are small ones: "Why that threshold?" "Why those exclusion criteria?" "Why this dataset?" Before your fair, walk your entire pipeline and confirm that every choice in it has a reason you can give in your own words. Good mentorship does not mean fewer of your fingerprints on the project. It means you understand every fingerprint, including the ones your mentor left.

Your opening 30 seconds

Every interview begins with some form of "tell me about your project." Script this one thing, not word for word, but beat by beat:

1. The problem, in plain language a judge from another field follows
2. What you did, one sentence
3. Your headline result, with the number
4. Why it matters, one sentence

No acronyms in the first thirty seconds. If the judge cannot follow your opening, the remaining fourteen minutes are spent recovering. If they can, they spend those minutes going deep, which is where prepared students win.

How to actually rehearse

Mock interviews, with someone technical enough to push. Not a run-through of your talk, an adversarial Q&A where they ask the four questions above plus every "why" in your methods. Three sessions is the difference between hearing the hard question for the first time from a judge and hearing it for the fifth time.

This is exactly what we run with YRI Fellows before their fairs: mock judging against the real rubric with full-time PhD mentors asking the questions judges ask, until the answers are automatic. Our Fellows do original research 1-on-1 with those same mentors, 87% publish their work, and 73% place at competitions, and the interview preparation is a large part of that last number.

If you want your project, and your answers, at that standard before your fair: yriscience.com.

Interview criteria summarized from the Regeneron ISEF Grand Award judging criteria published by Society for Science.