diff --git a/src/national_rail_timetable/mca_querying.py b/src/national_rail_timetable/mca_querying.py
new file mode 100644
index 0000000..2fd88dc
--- /dev/null
+++ b/src/national_rail_timetable/mca_querying.py
@@ -0,0 +1,191 @@
+"""
+MCA file querying to retrieve schedule(d service)s.
+"""
+# pyright: reportAny=false
+# pyright: reportUnknownVariableType=false
+# pyright: reportUnknownArgumentType=false
+# pyright: reportAttributeAccessIssue=false
+# pyright: reportOperatorIssue=false
+
+# Imports
+from dataclasses import dataclass
+from pathlib import Path
+from tempfile import TemporaryDirectory
+from typing import Callable, Self
+from zipfile import ZipFile
+import numpy as np
+from numpy.typing import NDArray
+from national_rail_timetable.nr_requests import fetch_nr_timetable_files
+from national_rail_timetable.mca_stubs import BS, BX, LI, LO, LT, CR
+
+
+# Functions
+def ux(s1: NDArray[np.byte]) -> NDArray[np.str_]:
+    x = s1.shape[1]
+    s1.dtype = f"S{x}"
+    return s1.astype(f"U{x}")[:, 0]
+
+
+def sx(s1: NDArray[np.byte]) -> NDArray[np.byte]:
+    x = s1.shape[1]
+    s1.dtype = f"S{x}"
+    return s1[:, 0]
+
+
+# Classes
+@dataclass
+class Timetable:
+    array: NDArray[np.byte]
+
+    @classmethod
+    def from_zipfile(
+        cls,
+        zipfile: ZipFile | None = None,
+    ) -> Self:
+        zipfile = zipfile if zipfile is not None else fetch_nr_timetable_files()
+        with TemporaryDirectory() as tempdir:
+            name = zipfile.extract(
+                [
+                    zipinfo
+                    for name, zipinfo in zipfile.NameToInfo.items()
+                    if name.split(".")[-1] == "MCA"
+                ][0],
+                path=tempdir,
+            )
+            array: NDArray[np.byte] = np.fromfile(
+                Path(tempdir) / name, dtype="S1"
+            ).reshape((-1, 82))[:, :-2]
+        return cls(array=array)
+
+    def bs_mask(self) -> NDArray[np.bool]:
+        return (self.array[:, 0] == b"B") & (self.array[:, 1] == b"S")
+
+    def lo_mask(self) -> NDArray[np.bool]:
+        return (self.array[:, 0] == b"L") & (self.array[:, 1] == b"O")
+
+    @property
+    def sns(self) -> NDArray[np.integer]:
+        return np.repeat(
+            (
+                sns := [
+                    0,
+                    *np.arange(self.array.shape[0])[self.bs_mask()],
+                    self.array.shape[0],
+                ]
+            )[:-1],
+            np.diff(sns),
+        )
+
+    def fetch_schedules(self, *sns: int) -> NDArray[np.byte]:
+        return self.array[np.isin(self.sns, sns)]
+
+
+@dataclass
+class Query:
+    tt: Timetable
+    sns: NDArray[np.integer] | None = None
+
+    def _query_from_mask(self, mask: NDArray[np.bool]) -> Query:
+        sns: NDArray[np.integer] = self.tt.sns[mask]
+        if self.sns is not None:
+            sns = np.intersect1d(sns, self.sns)
+        return Query(self.tt, sns)
+
+    def on_date(self, date: str) -> Query:
+        mask: NDArray[np.bool] = (
+            (sx(self.tt.array[:, LO().record_identity()]) == b"BS")
+            & (sx(self.tt.array[:, BS().date_runs_from()]) <= date.encode())
+            & (sx(self.tt.array[:, BS().date_runs_to()]) >= date.encode())
+        )
+        return self._query_from_mask(mask)
+
+    def origin(self, tiploc: str) -> Query:
+        mask: NDArray[np.bool] = (
+            (sx(self.tt.array[:, LO().record_identity()]) == b"LO")
+            & (
+                sx(self.tt.array[:, LO().location().start : LO().location().stop - 1])
+                == f"{tiploc:<7}".encode()
+            ),
+        )[0]
+        return self._query_from_mask(mask)
+
+    def dest(self, tiploc: str, call_number: str | int | None = None) -> Query:
+        mask: NDArray[np.bool] = (
+            (sx(self.tt.array[:, LO().record_identity()]) == b"LT")
+            & (
+                sx(self.tt.array[:, LT().location().start : LT().location().stop - 1])
+                == f"{tiploc:<7}".encode()
+            ),
+        )[0]
+        if call_number is not None:
+            mask &= (
+                sx(self.tt.array[:, LT().location().stop]) == str(call_number).encode()
+            )
+        return self._query_from_mask(mask)
+
+    def calls(self, tiploc: str, call_number: str | int | None = None) -> Query:
+        mask: NDArray[np.bool] = (
+            (sx(self.tt.array[:, LI().record_identity()]) == b"LI")
+            & (
+                sx(self.tt.array[:, LI().location().start : LI().location().stop - 1])
+                == f"{tiploc:<7}".encode()
+            ),
+        )[0]
+        if call_number is not None:
+            mask &= (
+                sx(self.tt.array[:, LI().location().stop]) == str(call_number).encode()
+            )
+        return self._query_from_mask(mask) | self.origin(tiploc) | self.dest(tiploc)
+
+    def get_field(self, mca_field: Callable[..., slice]):
+        record_type = str(mca_field).split("method ")[1].split(".")[0]
+        return sx(
+            self.result[sx(self.result[:, :2]) == record_type.encode(), mca_field()]
+        )
+
+    @property
+    def _a(self):
+        return self.tt.array
+
+    @property
+    def result(self) -> NDArray[np.byte]:
+        assert self.sns is not None
+        return self.tt.fetch_schedules(*self.sns)
+
+    def as_sx(self) -> NDArray[np.byte]:
+        return sx(self.result)
+
+    def as_ux(self) -> NDArray[np.str_]:
+        return ux(self.result)
+
+    def __and__(self, other: Query) -> Query:
+        assert self.tt is other.tt
+        assert self.sns is not None
+        assert other.sns is not None
+        return Query(self.tt, np.intersect1d(self.sns, other.sns))
+
+    def __or__(self, other: Query) -> Query:
+        assert self.tt is other.tt
+        assert self.sns is not None
+        assert other.sns is not None
+        return Query(self.tt, np.union1d(self.sns, other.sns))
+
+
+# Script
+def main():
+
+    try:
+        tt = Timetable(
+            np.load(Path(__file__).parents[2] / "data/cache.mca.npy"),
+        )
+    except FileNotFoundError:
+        tt = Timetable.from_zipfile()
+        np.save(Path(__file__).parents[2] / "data/cache.mca", tt.array)
+
+    result = Query(tt).on_date("260526").calls("CRMLNGT")
+    print(result.as_ux())
+    print(result.get_field(BX().retail_service_id))
+
+
+if __name__ == "__main__":
+    main()