diff --git a/src/analysis/diagnostic.rs b/src/analysis/diagnostic.rs
index 0c3be4e..9db39fa 100644
--- a/src/analysis/diagnostic.rs
+++ b/src/analysis/diagnostic.rs
@@ -172,6 +172,12 @@ pub struct StudentDiagnostic {
     /// One row per question, with no question in it.
     #[serde(skip_serializing_if = "Vec::is_empty")]
     pub questions: Vec<QuestionRow>,
+    /// Questions dropped from scoring, in number order.
+    ///
+    /// Reported separately from the question rows so the document can say once,
+    /// in prose, why the denominator is smaller than the paper was.
+    #[serde(skip_serializing_if = "Vec::is_empty")]
+    pub dropped_questions: Vec<u32>,
     /// Which lectures to go back to, the one that would repay the most time
     /// first.
     #[serde(skip_serializing_if = "Vec::is_empty")]
@@ -320,6 +326,13 @@ pub struct QuestionRow {
     /// Whether it was a bonus question.
     #[serde(skip_serializing_if = "std::ops::Not::not")]
     pub bonus: bool,
+    /// Whether it was dropped from scoring after the fact.
+    ///
+    /// A dropped question still appears in the map, because the student has the
+    /// paper in front of them and will look for it. What it must not do is
+    /// appear as an error they made.
+    #[serde(skip_serializing_if = "std::ops::Not::not")]
+    pub dropped: bool,
     /// Whether the student left it blank.
     pub blank: bool,
     /// The share of the class that answered it correctly, when comparison is on.
@@ -530,6 +543,14 @@ pub fn student(
         Vec::new()
     };
 
+    let mut dropped_questions: Vec<u32> = rows
+        .iter()
+        .filter(|r| r.dropped)
+        .map(|r| r.item_number)
+        .collect();
+    dropped_questions.sort_unstable();
+    dropped_questions.dedup();
+
     // Built from the response rows rather than from `questions`, so a report with
     // `--no-questions` still says where to go back to; it just does not name the
     // question numbers.
@@ -573,6 +594,7 @@ pub fn student(
         strengths,
         focus,
         questions,
+        dropped_questions,
         review_lectures,
         study,
     }
@@ -591,7 +613,10 @@ fn question_row(
     opts: &Options,
 ) -> QuestionRow {
     let blank = row.selected.is_empty() && row.eliminated.is_empty();
-    let missed = row.credit < 0.999;
+    // A dropped question cannot be missed. Without `counts()` here, a question
+    // thrown out after the exam still collects per-option feedback explaining an
+    // error the student is no longer being charged for.
+    let missed = row.credit < 0.999 && row.counts();
     let mut feedback = None;
     let mut hint = None;
     let mut misconception = None;
@@ -673,6 +698,7 @@ fn question_row(
         correct: row.correct,
         credit: row.credit,
         bonus: row.bonus,
+        dropped: row.dropped,
         blank,
         class_rate: opts
             .comparison
@@ -926,6 +952,11 @@ pub struct CohortDiagnostic {
     pub lectures: Vec<CohortLectureRow>,
     /// Per-question statistics.
     pub questions: Vec<CohortQuestionRow>,
+    /// Questions dropped from scoring, which are therefore absent from every
+    /// table above. Recorded so the report says why rather than leaving a gap in
+    /// the numbering.
+    #[serde(skip_serializing_if = "Vec::is_empty")]
+    pub dropped_questions: Vec<u32>,
     /// What to do about each question that raised something.
     pub triage: Triage,
     /// How the authored expectations did.
@@ -1478,6 +1509,13 @@ pub fn cohort(
         .filter_map(|item| questions.iter().find(|q| q.number == item.number).cloned())
         .collect();
 
+    let mut dropped_questions: Vec<u32> = set
+        .all_items()
+        .into_iter()
+        .filter(|n| set.item_dropped(*n))
+        .collect();
+    dropped_questions.sort_unstable();
+
     let default_options = course.policy.options_per_item;
     let triage = triage(&questions, threshold, default_options);
     let predictions = prediction_summary(analysis);
@@ -1490,6 +1528,7 @@ pub fn cohort(
         distribution: distribution(&percents),
         grades,
         lectures,
+        dropped_questions,
         triage,
         predictions,
         reliability: ReliabilityRow {
@@ -2193,6 +2232,7 @@ mod tests {
             strengths: Vec::new(),
             focus: Vec::new(),
             questions: Vec::new(),
+            dropped_questions: Vec::new(),
             review_lectures: Vec::new(),
             study: Vec::new(),
         })
diff --git a/src/export/typst/diagnostic.rs b/src/export/typst/diagnostic.rs
index 08ac1db..a321ad3 100644
--- a/src/export/typst/diagnostic.rs
+++ b/src/export/typst/diagnostic.rs
@@ -299,6 +299,7 @@ pub fn student_value(diagnostic: &StudentDiagnostic, config: &RenderConfig) -> V
                     value.insert_some("correct", question.correct.map(Value::Bool));
                     value.insert("credit", Value::Float(question.credit));
                     value.insert("bonus", Value::Bool(question.bonus));
+                    value.insert("dropped", Value::Bool(question.dropped));
                     value.insert("blank", Value::Bool(question.blank));
                     value.insert_some("class-rate", question.class_rate.map(Value::Float));
                     value.insert(
@@ -348,6 +349,17 @@ pub fn student_value(diagnostic: &StudentDiagnostic, config: &RenderConfig) -> V
         ),
     );
 
+    out.insert(
+        "dropped-questions",
+        Value::Array(
+            diagnostic
+                .dropped_questions
+                .iter()
+                .map(|n| Value::Int(*n as i64))
+                .collect(),
+        ),
+    );
+
     out.insert(
         "review-lectures",
         Value::Array(
@@ -612,6 +624,17 @@ pub fn cohort_value(diagnostic: &CohortDiagnostic, config: &RenderConfig) -> Val
         ),
     );
 
+    out.insert(
+        "dropped-questions",
+        Value::Array(
+            diagnostic
+                .dropped_questions
+                .iter()
+                .map(|n| Value::Int(*n as i64))
+                .collect(),
+        ),
+    );
+
     let triage_rows = |rows: &[crate::diagnostic::TriageRow]| -> Value {
         Value::Array(rows.iter().map(|row| triage_value(row, content)).collect())
     };
@@ -1128,6 +1151,7 @@ mod tests {
             gaps: Vec::new(),
             grades: Vec::new(),
             lectures: Vec::new(),
+            dropped_questions: Vec::new(),
             questions: Vec::new(),
             triage: crate::diagnostic::Triage::default(),
             predictions: crate::diagnostic::PredictionSummary::default(),
diff --git a/src/export/typst/templates/cohort-report.typ b/src/export/typst/templates/cohort-report.typ
index f0c5449..420ae8e 100644
--- a/src/export/typst/templates/cohort-report.typ
+++ b/src/export/typst/templates/cohort-report.typ
@@ -167,6 +167,7 @@
     band-hit: 11,
     biggest-surprise: (number: 14, expected: 0.45, observed: 0.86),
   ),
+  dropped-questions: (),
   revise: (),
   forms: (
     (id: "A", students: 12, mean: 73.5, sd: 10.2),
@@ -989,6 +990,17 @@
     One row per question, in the order they were numbered. Option letters are the
     bank's, not any one form's. See the guide above for what p, r, and D mean; the
     r column is colour-coded against the conventional cut points.
+    #{
+      let dropped = cb-data.at("dropped-questions", default: ())
+      if dropped.len() > 0 [
+        #plural(dropped.len(), "Question", "Questions")
+        #dropped.map(str).join(", ")
+        #plural(dropped.len(), "is", "are")
+        marked dropped in the assessment record, so
+        #plural(dropped.len(), "it is", "they are")
+        absent from this table and from every statistic above.
+      ]
+    }
   ]
 
   #table(
diff --git a/src/export/typst/templates/student-report.typ b/src/export/typst/templates/student-report.typ
index 237aae4..7a898c8 100644
--- a/src/export/typst/templates/student-report.typ
+++ b/src/export/typst/templates/student-report.typ
@@ -121,6 +121,7 @@
       correct: true,
       credit: 1.0,
       bonus: false,
+      dropped: false,
       blank: false,
       class-rate: 0.91,
       taught-in: (),
@@ -451,6 +452,26 @@
   },
 )
 
+#let dropped-questions = cb-data.at("dropped-questions", default: ())
+
+#if dropped-questions.len() > 0 [
+  #block(above: entry-gap)[
+    #pad(right: prose-pad)[
+      #text(size: size-lead)[
+        #plural(dropped-questions.len(), "Question", "Questions")
+        #dropped-questions.map(str).join(", ")
+        #plural(dropped-questions.len(), "was", "were")
+        dropped after the exam and #plural(dropped-questions.len(), "is", "are")
+        not part of anyone's score. Your percentage above is out of the
+        #plural(dropped-questions.len(), "remaining question", "remaining questions").
+        Nothing you wrote on
+        #plural(dropped-questions.len(), "it", "them")
+        counted for or against you.
+      ]
+    ]
+  ]
+]
+
 // ─────────────────────────────────────────────────────────────────────────────
 // Levels
 // ─────────────────────────────────────────────────────────────────────────────
@@ -750,7 +771,10 @@
 #let questions = cb-data.at("questions", default: ())
 
 #let question-box(q) = {
-  let color = if q.at("blank", default: false) { luma(160) } else if q.at("correct", default: false) == true {
+  let dropped = q.at("dropped", default: false)
+  let color = if dropped { luma(130) } else if q.at("blank", default: false) {
+    luma(160)
+  } else if q.at("correct", default: false) == true {
     ok-color
   } else if q.at("credit", default: 0.0) > 0.0 { mid-color } else { bad-color }
   box(
@@ -763,10 +787,15 @@
     #align(center)[
       #text(size: size-small, weight: "bold", fill: color.darken(18%))[#q.number]
       #{
-        let level = q.at("level", default: none)
-        if level != none {
+        if dropped {
           linebreak()
-          text(size: size-tag, fill: luma(120))[L#level]
+          text(size: size-tag, fill: luma(110))[out]
+        } else {
+          let level = q.at("level", default: none)
+          if level != none {
+            linebreak()
+            text(size: size-tag, fill: luma(120))[L#level]
+          }
         }
       }
     ]