diff --git a/e2e/binary/numfmt/BUILD.bazel b/e2e/binary/numfmt/BUILD.bazel
index 386ea98a852b36702508982ce6f98a7074bd2488..cc2898eaf6e2030f26468220061419b873472e8f 100644
--- a/e2e/binary/numfmt/BUILD.bazel
+++ b/e2e/binary/numfmt/BUILD.bazel
@@ -1,11 +1,30 @@
-load("@bazel_skylib//rules:build_test.bzl", "build_test")
+load("@bazel_skylib//rules:native_binary.bzl", "native_test")
+load("@rules_python_pytest//python_pytest:defs.bzl", "py_pytest_test")
 
-# TODO: write an _actual_ test for `numfmt`
+py_pytest_test(
+    name = "pytest",
+    size = "small",
+    srcs = ["format.py"],
+    data = [
+        "@ape//ape:numfmt",
+    ],
+    deps = [
+        "//binary:pytest",
+    ],
+)
 
-build_test(
-    name = "numfmt",
+native_test(
+    name = "version",
     size = "small",
-    tags = ["stub"],
-    targets = ["@ape//ape:numfmt"],
+    src = "@ape//ape:numfmt",
+    args = ["--version"],
+)
+
+test_suite(
+    name = "numfmt",
+    tests = [
+        "pytest",
+        "version",
+    ],
     visibility = ["//:__subpackages__"],
 )
diff --git a/e2e/binary/numfmt/format.py b/e2e/binary/numfmt/format.py
new file mode 100644
index 0000000000000000000000000000000000000000..2402011025d3bdbead5b609845f1b2be04a5eaec
--- /dev/null
+++ b/e2e/binary/numfmt/format.py
@@ -0,0 +1,13 @@
+from __future__ import annotations
+
+from subprocess import PIPE, run
+
+from binary import Tool
+
+
+def test_format(tool: Tool) -> None:
+    tool = tool("numfmt")
+    cmd = (tool, "--to=si", "1000")
+    r = run(cmd, stdout=PIPE, encoding="utf8")
+    assert r.returncode == 0, r.stdout
+    assert r.stdout.rstrip() == "1.0K"