How the Next Generation NCLEX Is Scored

If you trained on the older NCLEX, you learned a hard rule: on a select-all-that-apply question, you either got every option right or you got nothing. The Next Generation NCLEX (NGN), launched April 1, 2023, changed that. NCSBN now uses partial-credit scoring on items that have more than one correct answer, which means a response that is mostly right can earn points instead of zero.

This page explains what NCSBN has actually published about NGN scoring and separates it from what the test-prep industry has inferred. That distinction matters, because a lot of confidently stated "facts" about NGN scoring are not in any NCSBN document. Two current NCSBN sources cover the topic:

• The 2026 NCLEX-RN Test Plan. It is brief on this topic, stating only that "There is partial credit scoring for items for which more than one key exists. There are three methods for scoring items for partial credit: plus/minus, zero/one and rationale scoring."
• NCSBN's scoring-models FAQ. The test plan does not itself map item types to scoring rules, but it points candidates to the FAQs on NCLEX.com, and NCSBN maintains one there titled "What are the current scoring models?" It is live, actively maintained and current as of July 2026, and it names the item types under each rule. The table below marks every row it covers.

Four things are absent from that FAQ: highlight, drag-and-drop cloze, trend, and the point values attached to dyads and triads. Those trace to NCSBN's NGN-era briefs and educator webinars from 2019 through 2022, as relayed in nurse-educator scoring guides. They are consistent across sources and are the best available, but they are not current NCSBN candidate documentation, and the table's Source column labels them as such. Where a number or a rule is not published, this page says so rather than filling the gap.

Want to practice the formats? Try our Next Gen NCLEX practice questions or the complete guide to NGN question types.

Why NGN scoring moved away from all-or-nothing

The old scoring approach was dichotomous: every item was worth one point, and you earned it only by getting the entire item correct. On a select-all-that-apply item with five options, a candidate who identified four correct options and missed one scored identically to a candidate who understood nothing. That is a measurement problem, not just a fairness complaint. Two candidates with clearly different levels of clinical knowledge produced the same data point.

NGN scoring is polytomous, meaning a single item can yield a range of score points rather than just zero or one. NCSBN has described this as increasing the precision of measurement, because it allows the exam to distinguish between partial understanding and no understanding instead of collapsing both into an incorrect response.

There is a second reason tied to the exam's design. The NGN is built around the NCSBN Clinical Judgment Measurement Model, which breaks clinical judgment into six steps: recognize cues, analyze cues, prioritize hypotheses, generate solutions, take action, and evaluate outcomes. Clinical judgment is not naturally a yes-or-no skill, and item formats that ask you to sort, group, highlight, and prioritize produce responses with degrees of correctness. Partial credit lets the scoring match the format.

One thing that did not change: the NCLEX is still a computerized adaptive test. You are not scored on a percentage of questions answered correctly. The algorithm maintains a running estimate of your ability relative to the passing standard and re-estimates it after every item.

The three scoring models NCSBN uses

NCSBN names three partial-credit methods: zero/one, plus/minus, and rationale scoring.

Zero/one (0/1) scoring is the classic approach applied at the level of each scorable element. You earn one point for each correct response you provide and zero points for an incorrect one. Nothing is subtracted. The item's total score is the sum of your correct responses. This model is used where the number of responses you can give is constrained by the item's structure, such as when the item tells you exactly how many options to select or gives you a fixed number of targets to fill.

NCSBN's scoring FAQ is explicit that no penalty is implemented for selecting a distractor on matrix multiple choice or multiple response select N items, and that a bow-tie item always has a maximum of five points because it has five targets.

Plus/minus (+/-) scoring applies where you could over-respond or under-respond because the item does not tell you how many answers are correct. Each correct selection adds a point and each incorrect selection subtracts a point. Critically, the score is floored at zero: if your selections sum to a negative number, the score defaults to zero rather than going negative. You cannot damage your performance on other items by over-selecting on one item.

Where that zero floor is applied depends on the format:

• Select all that apply and highlight: every selection sits in one pool, so a wrong click can cancel a point you earned elsewhere in the same item.
• Matrix multiple response and multiple response grouping: NCSBN's scoring FAQ states that scores are summed across columns or rows and that all negative column or row scores are rounded to zero, with each column or row scored independently so that it does not negatively impact surrounding rows or columns. A wrong selection can only cancel credit inside its own column on a matrix multiple response item, or inside its own group on a grouping item.

Rationale scoring, sometimes called dependent scoring, applies to items that assess linked information where one part justifies another, as in a nurse should do X because of Y. These items appear as dyads, with two linked response targets, or triads, with three. To earn credit, the linked parts must be correct together; getting one half right while the other half is wrong does not earn partial credit for that pair.

NCSBN's current scoring FAQ confirms the pairing rule and nothing beyond it: you earn points when both responses in the pair are correct. It publishes no point values, and it does not use the words dyad or triad at all. The figures you will see quoted elsewhere, that a dyad is worth one point and a triad up to two points scored as linked pairs, come from NCSBN's NGN-era presentations and educator webinars rather than from current candidate documentation, so treat them as informed background, not a published rule. How partial pairs within a triad are credited has never been spelled out publicly.

How each NGN item type is scored

The Source column marks which rows NCSBN documents today and which rest on older NGN-era material.

Why each item type gets its scoring model

The mapping in the table follows from a single question: can you over-respond?

When the item constrains how much you can select, zero/one scoring applies. Bow-tie items give you exactly five targets to fill. Multiple response select N tells you the number to select. Cloze items have a fixed number of blanks, whether drop-down or drag-and-drop (the drag-and-drop placement rests on NGN-era materials). Drop-down table items have one drop-down per row. Matrix multiple choice items require exactly one selection per row. In all of these, you physically cannot select more than the intended number, so there is no need for a deterrent against over-selecting.

When the item lets you select freely, plus/minus scoring applies. Extended multiple response (select all that apply) does not tell you how many are correct. Matrix multiple response allows multiple checkboxes per row. Multiple response grouping allows one to all options within each group. Highlight items, both text and table, let you click as many or as few segments as you want (highlight's placement here rests on NGN-era materials, not the current FAQ). In each case, plus/minus scoring is what stops a candidate from selecting everything and collecting full credit by brute force.

Rationale scoring is defined by NCSBN's current FAQ in terms of what it measures rather than by format name: it applies to items assessing paired information, concepts that require justification through a rationale. The specific format labels you will see, drag-and-drop rationale and drop-down rationale, come from NGN-era materials. Either way, these items present a sentence with two or three linked targets.

Trend items are a special case. A trend item presents serial data such as vital signs or labs across multiple time points. Trend is best understood as a presentation style rather than a distinct scoring category, and it appears nowhere in NCSBN's current scoring FAQ. The scoring rule most likely follows whatever response format the item uses, so a trend item built as a matrix multiple response would be scored plus/minus, while one built as a drop-down cloze would be scored zero/one. That is an inference from the format, not a published NCSBN rule.

Case studies versus stand-alone items

The NGN delivers clinical judgment content two ways. A case study is a set of six items sharing one unfolding client presentation, with the six items mapped to the six steps of the Clinical Judgment Measurement Model. A stand-alone item is a single item that measures clinical judgment without a connected scenario.

NCSBN's FAQ defines a stand-alone item simply as one administered independent of other items, and states plainly that any type of item can be used as a stand-alone item. Bow-tie and trend were the formats showcased as stand-alone clinical judgment items in NCSBN's NGN rollout materials, and those materials present bow-tie as a stand-alone-only format that is not used inside case studies, a restriction NCSBN's current FAQ does not restate. But stand-alone describes where an item sits, not a short list of formats that qualify.

NCSBN also answers a question students ask constantly: whether missing the first item in a case study sinks the rest. It does not. NCSBN states that each item within a case study is independent of the others and scored accordingly, that candidates cannot go back to view previous responses, and that care is taken to avoid cuing within each case study.

On weighting, be careful. Each of the six items in a case study is scored on its own using whichever model fits its format. NCSBN has not published the specific psychometric weight that any individual item, case study or stand-alone, contributes to your ability estimate, and item contribution in a computerized adaptive test depends on item difficulty and information rather than a flat point value. Anyone telling you a case study is worth a specific multiple of a regular question is asserting something NCSBN has not published.

For overall structure, the 2026 NCLEX-RN Test Plan sets a minimum of 85 items and a maximum of 150, within a five-hour limit that includes all breaks. Of a minimum-length exam, 52 items come from the eight content areas at their specified percentages, 18 items make up three clinical judgment case studies, and the remaining 15 are unscored pretest items. The test plan also specifies approximately 10 percent stand-alone clinical judgment items, selected depending on exam length. Because clinical judgment is treated as an integrated process, case study items span content areas and are counted independently of the content-area-specific items. The 2026 NCLEX-PN Test Plan specifies the same exam length and composition; what differs between the RN and PN test plans is the content-category breakdown, not these structural figures.

Common misconceptions

Partial credit means guessing is now free. It is not. On plus/minus items, an incorrect selection actively subtracts. Guessing is genuinely free only on zero/one items, where nothing is deducted.

Every NGN item gives partial credit. No. NCSBN specifies partial credit for items for which more than one key exists. A single-answer multiple choice item is still one point or nothing, and NCSBN's current FAQ confirms that all single-response items are scored zero/one. Older NCSBN-derived materials extended that list to formats such as fill in the blank and ordered response, which the current FAQ does not enumerate. Those same older materials are worth treating carefully for another reason: they described traditional select all that apply as remaining zero/one, while NCSBN's current FAQ lists select all that apply under plus/minus scoring.

Bow-tie items are all or nothing. They are not. Bow-tie uses zero/one scoring across five separate targets, so it is one of the more forgiving formats. Missing one placement does not zero out the other four.

You can score negative points on a question and drag down your overall score. You cannot. Plus/minus item scores that sum below zero default to zero. The damage is contained within that item.

A short exam means you passed, and a long one means you failed. The 2026 test plan states directly that exam length is not an indication of a pass or fail result, and that a candidate may pass or fail regardless of length.

You need a specific percentage correct to pass. The NCLEX is not scored on percent correct. It is a computerized adaptive test that estimates ability relative to the passing standard, using the 95 percent confidence interval rule, the maximum-length rule, or the run-out-of-time rule to reach a decision.

You can tell which scoring model an item is using from a label on screen. You cannot. The exam does not announce the scoring rule. Format cues are your only signal, which is why recognizing the difference between constrained and unconstrained response formats is a usable skill.

What this means for your test day

Your response strategy should differ depending on which scoring model an item uses, and the format usually tells you.

• Read the format before you answer, and classify it. Ask one question: does the item limit how much I can select? If it names a number or gives you fixed targets or blanks, it is almost certainly zero/one, and unanswered targets are guaranteed zeros. If you can select freely, assume plus/minus and treat every selection as a bet.
• Fill every target on constrained items, even when you are unsure. On bow-tie, cloze, drop-down table, select N, and matrix multiple choice, a wrong answer costs nothing beyond the zero you would get for leaving it blank. Leaving a bow-tie target empty or skipping a cloze blank can only hurt.
• Apply a confidence threshold on plus/minus items. Before clicking an option on a select-all-that-apply, matrix multiple response, grouping, or highlight item, ask whether you could defend that choice to a preceptor. If the honest answer is no, leave it. Restraint is rewarded on these items in a way it never was under all-or-nothing scoring.
• Know what an uncertain click costs. On a select-all-that-apply or highlight item, if you are confident about three options and genuinely unsure about a fourth, selecting that fourth risks converting a three-point response into a two-point one. On matrix multiple response and grouping items that damage is contained, because NCSBN scores each column or row independently and floors it at zero. Omitting an option you should have selected costs you the point you did not earn, but it does not subtract from points you did earn.
• Resist over-highlighting. Highlight items tempt you to click every abnormal finding. Select only the segments that answer the specific question asked, because each irrelevant highlight subtracts.
• Evaluate rationale items as a unit, and read them backward. On dyad and triad items, pick the physiologic justification first, then confirm the action follows from it. Because the parts are scored together, a clinically true action attached to the wrong reason earns nothing for that pair.
• Treat each case study item as independent. Because each item within a case study is scored on its own, a wrong answer on the analyze cues item does not doom the take action item. You also cannot go back to a previous item. Reset and re-read the updated client data for each step.
• Ignore your item count as a performance signal. NCSBN states plainly that exam length does not indicate pass or fail. Monitoring your question number mid-exam costs focus and tells you nothing, and there is no percentage-correct target to chase because the exam is adaptive.

Partial credit is not a safety net that lowers the bar. It changes how your knowledge is measured, not what counts as competent. A response that is half right is still measured as half right, and the adaptive algorithm incorporates that into an ability estimate judged against the same passing standard that applies to every other item.

Frequently Asked Questions