diff --git a/cpp/ql/lib/change-notes/2026-10-15-bitwidth.md b/cpp/ql/lib/change-notes/2026-10-15-bitwidth.md
new file mode 100644
index 000000000000..7d1b3b4248c4
--- /dev/null
+++ b/cpp/ql/lib/change-notes/2026-10-15-bitwidth.md
@@ -0,0 +1,4 @@
+---
+category: feature
+---
+* Added a new predicate `getBitWidth` to the `Type` class, which yields the bit-width of the type.
diff --git a/cpp/ql/lib/experimental/semmle/code/cpp/rangeanalysis/extensions/ConstantShiftExprRange.qll b/cpp/ql/lib/experimental/semmle/code/cpp/rangeanalysis/extensions/ConstantShiftExprRange.qll
index 3f300d7aa8d6..0f7e1ce5ed40 100644
--- a/cpp/ql/lib/experimental/semmle/code/cpp/rangeanalysis/extensions/ConstantShiftExprRange.qll
+++ b/cpp/ql/lib/experimental/semmle/code/cpp/rangeanalysis/extensions/ConstantShiftExprRange.qll
@@ -21,7 +21,7 @@ bindingset[val]
 private predicate isValidShiftExprShift(float val, Expr l) {
   val >= 0 and
   // We use getFullyConverted because the spec says to use the *promoted* left operand
-  val < (l.getFullyConverted().getUnderlyingType().getSize() * 8)
+  val < l.getFullyConverted().getUnderlyingType().getBitWidth()
 }
 
 bindingset[val, shift, max_val]
diff --git a/cpp/ql/lib/semmle/code/cpp/Enum.qll b/cpp/ql/lib/semmle/code/cpp/Enum.qll
index 38263dacf7ac..6b30d7921157 100644
--- a/cpp/ql/lib/semmle/code/cpp/Enum.qll
+++ b/cpp/ql/lib/semmle/code/cpp/Enum.qll
@@ -55,6 +55,12 @@ class Enum extends UserType, IntegralOrEnumType {
     not exists(UserType.super.getSize()) and result = 4
   }
 
+  override int getBitWidth() {
+    result = UserType.super.getBitWidth()
+    or
+    not exists(UserType.super.getBitWidth()) and result = 32
+  }
+
   /** See `Type.isDeeplyConst` and `Type.isDeeplyConstBelow`. Internal. */
   override predicate isDeeplyConstBelow() { any() } // No subparts
 
diff --git a/cpp/ql/lib/semmle/code/cpp/Type.qll b/cpp/ql/lib/semmle/code/cpp/Type.qll
index 023aa1b6c3d4..78367dbf7be4 100644
--- a/cpp/ql/lib/semmle/code/cpp/Type.qll
+++ b/cpp/ql/lib/semmle/code/cpp/Type.qll
@@ -123,6 +123,16 @@ class Type extends Locatable, @type {
     usertypesize(underlyingElement(this), result, _)
   }
 
+  /**
+   * Gets the bit width of this type.
+   */
+  int getBitWidth() {
+    not bit_precise_int_types(underlyingElement(this), _) and
+    result = this.getSize() * 8
+    or
+    bit_precise_int_types(underlyingElement(this), result)
+  }
+
   /**
    * Gets the alignment of this type in bytes.
    */
@@ -1076,9 +1086,6 @@ class Char32Type extends IntegralType {
 class BitPreciseIntType extends IntegralType {
   BitPreciseIntType() { builtintypes(underlyingElement(this), _, [67, 68, 69], _, _, _) }
 
-  /** Get the bit-with of the bit-precise integer. */
-  int getBitWidth() { bit_precise_int_types(underlyingElement(this), result) }
-
   override string getAPrimaryQlClass() { result = "BitPreciseIntType" }
 }
 
@@ -1182,6 +1189,8 @@ class Decltype extends Type, NameQualifyingElement {
 
   override int getSize() { result = this.getBaseType().getSize() }
 
+  override int getBitWidth() { result = this.getBaseType().getBitWidth() }
+
   override int getAlignment() { result = this.getBaseType().getAlignment() }
 
   override int getPointerIndirectionLevel() {
@@ -1242,6 +1251,8 @@ class TypeofType extends Type {
 
   override int getSize() { result = this.getBaseType().getSize() }
 
+  override int getBitWidth() { result = this.getBaseType().getBitWidth() }
+
   override int getAlignment() { result = this.getBaseType().getAlignment() }
 
   override int getPointerIndirectionLevel() {
@@ -1387,6 +1398,8 @@ class IntrinsicTransformedType extends Type {
 
   override int getSize() { result = this.getBaseType().getSize() }
 
+  override int getBitWidth() { result = this.getBaseType().getBitWidth() }
+
   override int getAlignment() { result = this.getBaseType().getAlignment() }
 
   override int getPointerIndirectionLevel() {
@@ -1512,6 +1525,8 @@ class SpecifiedType extends DerivedType {
 
   override int getSize() { result = this.getBaseType().getSize() }
 
+  override int getBitWidth() { result = this.getBaseType().getBitWidth() }
+
   override int getAlignment() { result = this.getBaseType().getAlignment() }
 
   override int getPointerIndirectionLevel() {
@@ -1587,6 +1602,8 @@ class ArrayType extends DerivedType {
    */
   override int getSize() { result = this.getByteSize() }
 
+  override int getBitWidth() { result = this.getByteSize() * 8 }
+
   override string explain() {
     if exists(this.getArraySize())
     then
diff --git a/cpp/ql/lib/semmle/code/cpp/TypedefType.qll b/cpp/ql/lib/semmle/code/cpp/TypedefType.qll
index 69eb8f881d29..70aa933d6a2d 100644
--- a/cpp/ql/lib/semmle/code/cpp/TypedefType.qll
+++ b/cpp/ql/lib/semmle/code/cpp/TypedefType.qll
@@ -26,6 +26,8 @@ class TypedefType extends UserType {
 
   override int getSize() { result = this.getBaseType().getSize() }
 
+  override int getBitWidth() { result = this.getBaseType().getBitWidth() }
+
   override int getAlignment() { result = this.getBaseType().getAlignment() }
 
   override int getPointerIndirectionLevel() {
diff --git a/cpp/ql/lib/semmle/code/cpp/commons/Printf.qll b/cpp/ql/lib/semmle/code/cpp/commons/Printf.qll
index 624465761c2c..dfda2f4796c5 100644
--- a/cpp/ql/lib/semmle/code/cpp/commons/Printf.qll
+++ b/cpp/ql/lib/semmle/code/cpp/commons/Printf.qll
@@ -1193,7 +1193,7 @@ class FormatLiteral extends Literal instanceof StringLiteral {
           // Subtract one in the exponent because one bit is for the sign.
           // Add 1 to account for the possible sign in the output.
           typeBasedBound =
-            1 + lengthInBase10(2.pow(this.getIntegralDisplayType(n).getSize() * 8 - 1)) and
+            1 + lengthInBase10(2.pow(this.getIntegralDisplayType(n).getBitWidth() - 1)) and
           // The second case uses range analysis to deduce a length that's shorter than the length
           // of the number -2^31.
           exists(Expr arg, float lower, float upper |
@@ -1227,7 +1227,7 @@ class FormatLiteral extends Literal instanceof StringLiteral {
         // e.g. 2^32 - 1 = "4294967295"
         exists(float typeBasedBound, float valueBasedBound |
           // The first case handles length sub-specifiers
-          typeBasedBound = lengthInBase10(2.pow(this.getIntegralDisplayType(n).getSize() * 8) - 1) and
+          typeBasedBound = lengthInBase10(2.pow(this.getIntegralDisplayType(n).getBitWidth()) - 1) and
           // The second case uses range analysis to deduce a length that's shorter than
           // the length of the number 2^31 - 1.
           exists(Expr arg, float lower, float upper |
@@ -1239,7 +1239,7 @@ class FormatLiteral extends Literal instanceof StringLiteral {
               lengthInBase10(max(float cand |
                   // If lower can be negative we use `(unsigned)-1` as the candidate value.
                   lower < 0 and
-                  cand = 2.pow(any(IntType t | t.isUnsigned()).getSize() * 8)
+                  cand = 2.pow(any(IntType t | t.isUnsigned()).getBitWidth())
                   or
                   cand = upper
                 )) and
@@ -1275,7 +1275,7 @@ class FormatLiteral extends Literal instanceof StringLiteral {
               lengthInBase16(max(float cand |
                   // If lower can be negative we use `(unsigned)-1` as the candidate value.
                   lower < 0 and
-                  cand = 2.pow(any(IntType t | t.isUnsigned()).getSize() * 8)
+                  cand = 2.pow(any(IntType t | t.isUnsigned()).getBitWidth())
                   or
                   cand = upper
                 )) and
@@ -1304,12 +1304,12 @@ class FormatLiteral extends Literal instanceof StringLiteral {
         exists(int sizeBits, int baseLen |
           sizeBits =
             min(int bits |
-              bits = this.getIntegralDisplayType(n).getSize() * 8
+              bits = this.getIntegralDisplayType(n).getBitWidth()
               or
               exists(IntegralType t |
                 t = this.getUse().getConversionArgument(n).getType().getUnderlyingType()
               |
-                t.isUnsigned() and bits = t.getSize() * 8
+                t.isUnsigned() and bits = t.getBitWidth()
               )
             ) and
           baseLen = (sizeBits / 3.0).ceil() and
diff --git a/cpp/ql/lib/semmle/code/cpp/controlflow/internal/ConstantExprs.qll b/cpp/ql/lib/semmle/code/cpp/controlflow/internal/ConstantExprs.qll
index 84e96fa9c468..5329fb92a5a4 100644
--- a/cpp/ql/lib/semmle/code/cpp/controlflow/internal/ConstantExprs.qll
+++ b/cpp/ql/lib/semmle/code/cpp/controlflow/internal/ConstantExprs.qll
@@ -344,12 +344,12 @@ private int convertIntToType(int val, IntegralType t) {
   then if val = 0 then result = 0 else result = 1
   else
     if t.isUnsigned()
-    then val >= 0 and val.bitShiftRight(t.getSize() * 8) = 0 and result = val
+    then val >= 0 and val.bitShiftRight(t.getBitWidth()) = 0 and result = val
     else
-      if val >= 0 and val.bitShiftRight(t.getSize() * 8 - 1) = 0
+      if val >= 0 and val.bitShiftRight(t.getBitWidth() - 1) = 0
       then result = val
       else (
-        (-(val + 1)).bitShiftRight(t.getSize() * 8 - 1) = 0 and result = val
+        (-(val + 1)).bitShiftRight(t.getBitWidth() - 1) = 0 and result = val
       )
 }
 
diff --git a/cpp/ql/lib/semmle/code/cpp/rangeanalysis/RangeAnalysisUtils.qll b/cpp/ql/lib/semmle/code/cpp/rangeanalysis/RangeAnalysisUtils.qll
index 03dbc56dc99c..3b316f5bce5c 100644
--- a/cpp/ql/lib/semmle/code/cpp/rangeanalysis/RangeAnalysisUtils.qll
+++ b/cpp/ql/lib/semmle/code/cpp/rangeanalysis/RangeAnalysisUtils.qll
@@ -406,7 +406,7 @@ predicate cmpWithLinearBound(
  */
 private predicate typeBounds0(ArithmeticType t, float lb, float ub) {
   exists(IntegralType integralType, float limit |
-    integralType = t and limit = 2.pow(8 * integralType.getSize())
+    integralType = t and limit = 2.pow(integralType.getBitWidth())
   |
     if integralType instanceof BoolType
     then lb = 0 and ub = 1
@@ -438,7 +438,7 @@ private IntegralType getUnderlyingTypeForEnum(Enum e) {
   not e.hasExplicitUnderlyingType() and
   result.isSigned() and
   exists(IntType intType |
-    if max(e.getAnEnumConstant().getValue().toFloat()) >= 2.pow(8 * intType.getSize() - 1)
+    if max(e.getAnEnumConstant().getValue().toFloat()) >= 2.pow(intType.getBitWidth() - 1)
     then result instanceof LongType
     else result = intType
   )
diff --git a/cpp/ql/src/Likely Bugs/AmbiguouslySignedBitField.ql b/cpp/ql/src/Likely Bugs/AmbiguouslySignedBitField.ql
index 1fabf449ab35..3244d6748e9f 100644
--- a/cpp/ql/src/Likely Bugs/AmbiguouslySignedBitField.ql	
+++ b/cpp/ql/src/Likely Bugs/AmbiguouslySignedBitField.ql	
@@ -29,7 +29,7 @@ where
   // GLib's gboolean is a typedef for a type representing BoolType.
   not bf.getType().hasName("gboolean") and
   // If this is true, then there cannot be unsigned sign extension or overflow.
-  not bf.getDeclaredNumBits() = bf.getType().getSize() * 8 and
+  not bf.getDeclaredNumBits() = bf.getType().getBitWidth() and
   not bf.isAnonymous() and
   not bf.isFromUninstantiatedTemplate(_)
 select bf,
diff --git a/cpp/ql/src/Likely Bugs/Conversion/ConversionChangesSign.ql b/cpp/ql/src/Likely Bugs/Conversion/ConversionChangesSign.ql
index c89c71fb50b7..35843512ac05 100644
--- a/cpp/ql/src/Likely Bugs/Conversion/ConversionChangesSign.ql	
+++ b/cpp/ql/src/Likely Bugs/Conversion/ConversionChangesSign.ql	
@@ -23,7 +23,7 @@ where
   not (
     e1.isConstant() and
     0 <= e1.getValue().toInt() and
-    e1.getValue().toInt() <= ((it2.getSize() * 8 - 1) * 2.log()).exp()
+    e1.getValue().toInt() <= ((it2.getBitWidth() - 1) * 2.log()).exp()
   ) and
   not e1.isConstant()
 select e1,
diff --git a/cpp/ql/src/Likely Bugs/Conversion/ImplicitDowncastFromBitfield.ql b/cpp/ql/src/Likely Bugs/Conversion/ImplicitDowncastFromBitfield.ql
index 237d20c8f628..5f883e651ddb 100644
--- a/cpp/ql/src/Likely Bugs/Conversion/ImplicitDowncastFromBitfield.ql	
+++ b/cpp/ql/src/Likely Bugs/Conversion/ImplicitDowncastFromBitfield.ql	
@@ -20,8 +20,8 @@ where
     then fct = va.getFullyConverted().getType().(ReferenceType).getBaseType()
     else fct = va.getFullyConverted().getType()
   ) and
-  fi.getNumBits() > fct.getSize() * 8 and
-  va.getExplicitlyConverted().getType().getSize() > fct.getSize() and
+  fi.getNumBits() > fct.getBitWidth() and
+  va.getExplicitlyConverted().getType().getBitWidth() > fct.getBitWidth() and
   va.getTarget() = fi and
   not fct.getUnspecifiedType() instanceof BoolType
 select va, "Implicit downcast of bitfield $@.", fi, fi.toString()
diff --git a/cpp/ql/src/Likely Bugs/Memory Management/Padding/Suboptimal64BitType.ql b/cpp/ql/src/Likely Bugs/Memory Management/Padding/Suboptimal64BitType.ql
index f2cf4367d86d..5393c17b3ea3 100644
--- a/cpp/ql/src/Likely Bugs/Memory Management/Padding/Suboptimal64BitType.ql	
+++ b/cpp/ql/src/Likely Bugs/Memory Management/Padding/Suboptimal64BitType.ql	
@@ -14,7 +14,7 @@ from
   PaddedType t, Architecture arch, WideCharType wc, int holes, int size, int percentage, int optimum
 where
   arch.pointerSize() = 64 and // Select 64-bit architecture
-  arch.wideCharSize() = (wc.getSize() * 8) and // Select Windows(sizeof(wchar_t == 2)) or non-Windows(sizeof(wchar_t == 4))
+  arch.wideCharSize() = wc.getBitWidth() and // Select Windows(sizeof(wchar_t == 2)) or non-Windows(sizeof(wchar_t == 4))
   t.isPrecise() and
   optimum = t.optimalSize(arch) and
   size = arch.paddedSize(t) and
diff --git a/cpp/ql/src/jsf/4.20 Unions and Bit Fields/AV Rule 154.ql b/cpp/ql/src/jsf/4.20 Unions and Bit Fields/AV Rule 154.ql
index cd08362ad35a..87086db9e9a9 100644
--- a/cpp/ql/src/jsf/4.20 Unions and Bit Fields/AV Rule 154.ql	
+++ b/cpp/ql/src/jsf/4.20 Unions and Bit Fields/AV Rule 154.ql	
@@ -22,7 +22,7 @@ where
   not bf.getUnderlyingType() instanceof Enum and
   not bf.getUnspecifiedType() instanceof BoolType and
   not bf.getType().hasName("BOOL") and // At least for C programs on Windows, BOOL is a common typedef for a type representing BoolType.
-  not bf.getDeclaredNumBits() = bf.getType().getSize() * 8 and // If this is true, then there cannot be unsigned sign extension or overflow.
+  not bf.getDeclaredNumBits() = bf.getType().getBitWidth() and // If this is true, then there cannot be unsigned sign extension or overflow.
   not bf.isAnonymous()
 select bf,
   "Bit field " + bf.getName() + " of type " + bf.getUnderlyingType().getName() +
diff --git a/cpp/ql/src/jsf/4.23 Type Conversions/AV Rule 180.ql b/cpp/ql/src/jsf/4.23 Type Conversions/AV Rule 180.ql
index 16727add3fff..7e03caef0b91 100644
--- a/cpp/ql/src/jsf/4.23 Type Conversions/AV Rule 180.ql	
+++ b/cpp/ql/src/jsf/4.23 Type Conversions/AV Rule 180.ql	
@@ -61,10 +61,10 @@ class ImplicitConversion extends Expr {
   int getSourceSize() {
     if this instanceof IntConstantExpr
     then result = this.(IntConstantExpr).getRequiredSizeBits()
-    else result = this.getUnderlyingSourceType().getSize() * 8
+    else result = this.getUnderlyingSourceType().getBitWidth()
   }
 
-  int getTargetSize() { result = this.getUnderlyingTargetType().getSize() * 8 }
+  int getTargetSize() { result = this.getUnderlyingTargetType().getBitWidth() }
 
   predicate isSourceSignedInt() {
     if this instanceof IntConstantExpr