diff --git a/CHANGELOG.md b/CHANGELOG.md
index ad1a945..dc35c96 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -8,6 +8,7 @@
 
 ### Fixed
 
+- Fixed `FileStore` losing records when configured with a single segment: after the first compaction new writes went to a segment outside the configured range and were dropped on reload. `init()` now uses a minimum of two segments.
 - Fixed `FileStore` compaction running with the active segment still open, which LittleFS and FatFS refuse to delete, and fixed the store being left without a writable segment (or with a stray out-of-range segment file) after a failed compaction (#6, #11).
 - Fixed compaction recovery paths that could lose data: a power loss after the compacted segment was moved into place but before the journal was cleared deleted all data on the next boot; a failed rename while finalizing deleted all data and still reported success; recovery from a compaction interrupted mid-way kept a stale index, and a later compaction attempt could truncate the only copy of the interrupted compaction's records. A compaction interrupted by a write failure is now recovered immediately rather than at the next boot.
 
diff --git a/docs/microstore_user_guide.md b/docs/microstore_user_guide.md
index f821e13..6ea024d 100644
--- a/docs/microstore_user_guide.md
+++ b/docs/microstore_user_guide.md
@@ -241,7 +241,7 @@ if (!store.init(fs, "/cfg")) {
 Optional `init()` arguments:
 
 - `clearOnInit` — wipe everything before opening (useful for tests).
-- `segment_size`, `segment_count` — override the defaults at runtime (still bounded by the compile-time `USTORE_MAX_*` macros).
+- `segment_size`, `segment_count` — override the defaults at runtime (still bounded by the compile-time `USTORE_MAX_*` macros). `segment_count` has a minimum of 2: compaction writes into seg0 and new records then go to seg1, so a smaller value is raised to 2.
 
 Always call `store.close()` before destruction if you want a clean flush; the destructor does flush, but `close()` also closes the index and active segment explicitly.
 
diff --git a/include/microStore/FileStore.h b/include/microStore/FileStore.h
index 264b275..01af36b 100644
--- a/include/microStore/FileStore.h
+++ b/include/microStore/FileStore.h
@@ -142,6 +142,14 @@ class BasicFileStore
 	{
 		if (segment_size > 0) _segment_size = segment_size;
 		if (segment_count > 0) _segment_count = segment_count;
+		// Compaction writes the live records into seg0 and new writes then go to
+		// seg1, so a store needs at least two segments. With one, every write
+		// after the first compaction lands in seg1, outside the range that
+		// compaction and index rebuilds scan, and is lost on the next reload.
+		if (_segment_count < 2) {
+			USTORE_LOG("[ustore] init: segment_count %u is below the minimum of 2, using 2\n", _segment_count);
+			_segment_count = 2;
+		}
 		USTORE_LOG("[ustore] init: Initializing FileStore with prefix=%s, segment_size=%u, segment_count=%u\n", prefix, _segment_size, _segment_count);
 
 		_filesystem = filesystem;
diff --git a/test/test_file_store/test_file_store.cpp b/test/test_file_store/test_file_store.cpp
index f8ce6f4..1d929b8 100644
--- a/test/test_file_store/test_file_store.cpp
+++ b/test/test_file_store/test_file_store.cpp
@@ -1104,6 +1104,41 @@ void test_dir_prefix_compact() {
 void setUp()    {}
 void tearDown() {}
 
+// A store needs two segments: compaction fills seg0 and new writes go to seg1.
+// Asked for one, init() must use two, or every write after the first
+// compaction lands outside the scanned range and is lost on reload.
+void test_file_store_single_segment_count_keeps_data() {
+    reset_ram_fs();
+
+    uint32_t now = microStore::time() - 10;
+    char k[8], v[32];
+    {
+        microStore::FileStore store;
+        auto fs = make_ram_fs();
+        store.init(fs, "/p", /*clearOnInit=*/false, /*segment_size=*/256, /*segment_count=*/1);
+        // Overwrite ten keys six times: enough to fill a 256-byte segment
+        // several times over and force repeated compaction.
+        for (int i = 0; i < 60; i++) {
+            snprintf(k, sizeof(k), "k%02d", i % 10);
+            snprintf(v, sizeof(v), "value-%02d-xxxxxxxx", i);
+            TEST_ASSERT_TRUE(store.put(k, v, /*ttl=*/0, now));
+        }
+    }
+
+    microStore::FileStore reader;
+    auto fs = make_ram_fs();
+    reader.init(fs, "/p", /*clearOnInit=*/false, /*segment_size=*/256, /*segment_count=*/1);
+    TEST_ASSERT_EQUAL(10u, reader.size());
+    for (int i = 0; i < 10; i++) {
+        snprintf(k, sizeof(k), "k%02d", i);
+        snprintf(v, sizeof(v), "value-%02d-xxxxxxxx", 50 + i);
+        uint8_t buf[64]; uint16_t sz = sizeof(buf);
+        TEST_ASSERT_TRUE(reader.get(k, buf, &sz));
+        TEST_ASSERT_EQUAL(strlen(v), sz);
+        TEST_ASSERT_EQUAL(0, memcmp(buf, v, sz));
+    }
+}
+
 int runUnityTests(void) {
     UNITY_BEGIN();
     // TTL tests
@@ -1127,6 +1162,7 @@ int runUnityTests(void) {
     // Additional edge-case tests
     RUN_TEST(test_file_store_ttl_exists_expires);
     RUN_TEST(test_file_store_compact_basic);
+    RUN_TEST(test_file_store_single_segment_count_keeps_data);
     RUN_TEST(test_file_store_compact_keeps_store_writable);
     RUN_TEST(test_file_store_compact_failure_keeps_store_writable);
     RUN_TEST(test_file_store_threshold_compact_failure_keeps_store_writable);