diff --git a/src/freshdata/context/validate.py b/src/freshdata/context/validate.py
index 390c5a9..b7016d2 100644
--- a/src/freshdata/context/validate.py
+++ b/src/freshdata/context/validate.py
@@ -7,7 +7,8 @@
 
 from __future__ import annotations
 
-from typing import TYPE_CHECKING
+import math
+from typing import TYPE_CHECKING, Any
 
 from ..findings import FindingList, QualityFinding
 from .compiler import compile_context, effective_columns, resolve_policy
@@ -65,15 +66,92 @@ def _check_unique(series: pd.Series, c: ColumnConstraint) -> QualityFinding | No
     )
 
 
+#: Case-insensitive spellings accepted for a boolean column's allowed values.
+_BOOL_WORDS = {
+    "true": True,
+    "t": True,
+    "yes": True,
+    "y": True,
+    "1": True,
+    "false": False,
+    "f": False,
+    "no": False,
+    "n": False,
+    "0": False,
+}
+
+
+def _as_number(value: object) -> int | float | None:
+    """*value* as a finite Python number, or ``None`` when it is not numeric."""
+    if isinstance(value, bool):
+        return None
+    if isinstance(value, (int, float)):
+        number: int | float = value
+    else:
+        text = str(value).strip()
+        try:
+            number = int(text)
+        except ValueError:
+            try:
+                number = float(text)
+            except ValueError:
+                return None
+    return number if math.isfinite(number) else None
+
+
+def _as_bool(value: object) -> bool | None:
+    """*value* as a bool (true/false/yes/no/1/0/t/f/y/n, any case), else ``None``."""
+    if isinstance(value, bool):
+        return value
+    return _BOOL_WORDS.get(str(value).strip().lower())
+
+
+def _dedupe(values: list[Any]) -> list[Any]:
+    out: list[Any] = []
+    for v in values:
+        if not any(type(v) is type(o) and v == o for o in out):
+            out.append(v)
+    return out
+
+
+def _typed_allowed(series: pd.Series, raw: list[Any]) -> tuple[list[Any], list[Any], pd.Series]:
+    """Return ``(value_set, comparable, observed)`` for the column's dtype.
+
+    * boolean columns (``bool`` / nullable ``boolean``): allowed entries map
+      case-insensitively via :data:`_BOOL_WORDS`;
+    * other numeric columns: allowed entries are parsed as numbers and compared
+      with exact numeric equality (so ``"1"`` matches ``1.0``);
+    * everything else: today's string comparison, unchanged.
+
+    ``value_set`` keeps entries that do not convert as their original string so
+    nothing declared is dropped from the exported set; ``comparable`` holds only
+    the converted entries, since an unconvertible one can never match a value of
+    that dtype. Missing values are excluded from ``observed`` in every case.
+    """
+    from pandas.api.types import is_bool_dtype, is_numeric_dtype  # noqa: PLC0415
+
+    observed = series.dropna()
+    if is_bool_dtype(series.dtype) or is_numeric_dtype(series.dtype):
+        convert = _as_bool if is_bool_dtype(series.dtype) else _as_number
+        converted = [convert(v) for v in raw]
+        comparable = _dedupe([v for v in converted if v is not None])
+        value_set = _dedupe([str(v) if t is None else t for v, t in zip(raw, converted)])
+        if is_bool_dtype(series.dtype):
+            observed = observed.astype(bool)
+        return value_set, comparable, observed
+    values = [str(v) for v in raw]
+    return values, values, observed.astype(str)
+
+
 def _check_allowed_values(series: pd.Series, c: ColumnConstraint) -> QualityFinding | None:
-    values = [str(v) for v in c.params.get("values", ())]
-    if not values:
+    raw = list(c.params.get("values", ()))
+    if not raw:
         return None
-    observed = series.dropna().astype(str)
-    bad = observed[~observed.isin(values)]
+    values, comparable, observed = _typed_allowed(series, raw)
+    bad = observed[~observed.isin(comparable)]
     if bad.empty:
         return None
-    examples = sorted(set(bad))[:5]
+    examples = sorted(set(bad.tolist()))[:5]
     return QualityFinding.create(
         severity="error",
         step=_STEP,
diff --git a/tests/context/test_allowed_values_typed.py b/tests/context/test_allowed_values_typed.py
new file mode 100644
index 0000000..dd680e2
--- /dev/null
+++ b/tests/context/test_allowed_values_typed.py
@@ -0,0 +1,123 @@
+"""``allowed_values`` compares by column dtype for numeric and boolean columns (#255)."""
+
+from __future__ import annotations
+
+import json
+import warnings
+
+import pandas as pd
+import pytest
+
+import freshdata as fd
+
+
+def _allowed(findings):
+    return {f.column: f for f in findings if f.rule_name == "context.allowed_values"}
+
+
+def _validate(frame, context):
+    with warnings.catch_warnings():
+        warnings.simplefilter("ignore")
+        return fd.validate(frame, context=context)
+
+
+def test_issue_repro_float_and_bool_columns_pass():
+    frame = pd.DataFrame({"rating": [1.0, 2.0, 3.0, None], "active": [True, False, True, False]})
+    findings = _validate(
+        frame, "Allowed rating values are 1, 2, 3. Allowed active values are true, false."
+    )
+    assert _allowed(findings) == {}
+
+
+def test_float_column_with_decimal_allowed_values():
+    frame = pd.DataFrame({"score": [1.5, 2.5, 1.5, None]})
+    assert _allowed(_validate(frame, "Allowed score values are 1.5, 2.5.")) == {}
+
+
+def test_float_column_violation_reports_typed_values():
+    frame = pd.DataFrame({"score": [1.5, 2.5, 3.5, None]})
+    finding = _allowed(_validate(frame, "Allowed score values are 1.5, 2.5."))["score"]
+    assert finding.extra["value_set"] == [1.5, 2.5]
+    assert finding.extra["n_violations"] == 1
+    assert finding.observed_value == [3.5]
+
+
+def test_int_column_exact_numeric_equality():
+    frame = pd.DataFrame({"grade": [1, 2, 3, 4]})
+    finding = _allowed(_validate(frame, "Allowed grade values are 1, 2, 3."))["grade"]
+    assert finding.extra["value_set"] == [1, 2, 3]
+    assert all(type(v) is int for v in finding.extra["value_set"])
+    assert finding.extra["n_violations"] == 1
+    assert finding.observed_value == [4]
+
+
+def test_nullable_int_column_ignores_missing():
+    frame = pd.DataFrame({"grade": pd.array([1, None, 2], dtype="Int64")})
+    assert _allowed(_validate(frame, "Allowed grade values are 1, 2.")) == {}
+
+
+def test_numeric_column_non_numeric_entries_cannot_match():
+    frame = pd.DataFrame({"grade": [1, 2]})
+    finding = _allowed(_validate(frame, "Allowed grade values are 1, low."))["grade"]
+    assert finding.extra["value_set"] == [1, "low"]
+    assert finding.extra["n_violations"] == 1
+    assert finding.observed_value == [2]
+
+
+@pytest.mark.parametrize("words", ["true, false", "yes, no", "TRUE, False", "1, 0", "Y, n"])
+def test_bool_column_word_spellings(words):
+    frame = pd.DataFrame({"active": [True, False, True]})
+    findings = _validate(frame, f"Allowed active values are {words}.")
+    assert _allowed(findings) == {}
+
+
+def test_bool_column_violation_reports_typed_values():
+    frame = pd.DataFrame({"active": [True, False, True]})
+    finding = _allowed(_validate(frame, "Allowed active values are yes."))["active"]
+    assert finding.extra["value_set"] == [True]
+    assert finding.extra["n_violations"] == 1
+    assert finding.observed_value == [False]
+
+
+def test_nullable_boolean_column():
+    frame = pd.DataFrame({"active": pd.array([True, None, False], dtype="boolean")})
+    assert _allowed(_validate(frame, "Allowed active values are true, false.")) == {}
+    finding = _allowed(_validate(frame, "Allowed active values are true."))["active"]
+    assert finding.extra["value_set"] == [True]
+    assert finding.extra["n_violations"] == 1
+
+
+def test_string_column_behaviour_unchanged():
+    frame = pd.DataFrame({"status": ["active", "zombie", "True", "1.0", None]})
+    finding = _allowed(_validate(frame, "Allowed status values are active, true, 1."))["status"]
+    assert finding.extra["value_set"] == ["active", "true", "1"]
+    assert finding.extra["n_violations"] == 3
+    assert finding.observed_value == ["1.0", "True", "zombie"]
+
+
+def test_gx_export_uses_typed_value_set(tmp_path):
+    frame = pd.DataFrame({"rating": [1.0, 2.0, 9.0], "active": [True, False, True]})
+    findings = _validate(
+        frame, "Allowed rating values are 1, 2, 3. Allowed active values are true."
+    )
+    path = tmp_path / "suite.json"
+    suite = fd.export_gx_suite(findings, "s", str(path))
+    value_sets = {
+        e["kwargs"]["column"]: e["kwargs"]["value_set"]
+        for e in suite["expectations"]
+        if "value_set" in e["kwargs"]
+    }
+    assert value_sets == {"rating": [1, 2, 3], "active": [True]}
+    on_disk = json.loads(path.read_text(encoding="utf-8"))
+    assert on_disk == suite
+
+
+def test_dbt_export_emits_typed_scalars(tmp_path):
+    frame = pd.DataFrame({"rating": [1.0, 9.0], "active": [True, False]})
+    findings = _validate(
+        frame, "Allowed rating values are 1, 2.5. Allowed active values are true."
+    )
+    text = fd.export_dbt_tests(findings, "m", str(tmp_path / "schema.yml"))
+    assert "- 1\n" in text and "- 2.5\n" in text
+    assert "- true\n" in text
+    assert '"1"' not in text and '"true"' not in text