diff --git a/src/uu/od/src/parse_formats.rs b/src/uu/od/src/parse_formats.rs
index 1103b77cac3..e3312b87603 100644
--- a/src/uu/od/src/parse_formats.rs
+++ b/src/uu/od/src/parse_formats.rs
@@ -86,7 +86,8 @@ fn od_format_type(type_char: FormatType, byte_size: u8) -> Option<FormatterItemI
         (FormatType::HexadecimalInt, 8) => Some(FORMAT_ITEM_HEX64),
 
         (FormatType::Float, 2) => Some(FORMAT_ITEM_F16),
-        (FormatType::Float, 0 | 4) => Some(FORMAT_ITEM_F32),
+        (FormatType::Float, 4) => Some(FORMAT_ITEM_F32),
+        (FormatType::Float, 0) => Some(FORMAT_ITEM_F64),
         (FormatType::Float, 8) => Some(FORMAT_ITEM_F64),
         (FormatType::Float, 16) => Some(FORMAT_ITEM_LONG_DOUBLE),
 
@@ -521,7 +522,7 @@ fn test_long_format_x_default() {
 fn test_long_format_f_default() {
     assert_eq!(
         parse_format_flags_str(&["od", "--format=f"]).unwrap(),
-        vec![FORMAT_ITEM_F32]
+        vec![FORMAT_ITEM_F64]
     );
 }
 
@@ -587,7 +588,7 @@ fn test_mixed_formats() {
             ParsedFormatterItemInfo::new(FORMAT_ITEM_HEX8, false),   // tx1
             ParsedFormatterItemInfo::new(FORMAT_ITEM_DEC16U, false), // tu2
             ParsedFormatterItemInfo::new(FORMAT_ITEM_C, false),      // tc
-            ParsedFormatterItemInfo::new(FORMAT_ITEM_F32, false),    // tf
+            ParsedFormatterItemInfo::new(FORMAT_ITEM_F64, false),    // tf
             ParsedFormatterItemInfo::new(FORMAT_ITEM_HEX16, false),  // x
         ]
     );
diff --git a/tests/by-util/test_od.rs b/tests/by-util/test_od.rs
index 21cef8404ea..823f2d1dc51 100644
--- a/tests/by-util/test_od.rs
+++ b/tests/by-util/test_od.rs
@@ -223,6 +223,52 @@ fn test_bfloat16_compact() {
         .stdout_only("               1               1\n");
 }
 
+#[test]
+fn test_tf_default_is_double() {
+    let input: [u8; 8] = [0x00, 0x00, 0x80, 0x3f, 0x00, 0x00, 0x00, 0x40];
+    let default_output = new_ucmd!()
+        .arg("--endian=little")
+        .arg("-An")
+        .arg("-tf")
+        .run_piped_stdin(&input[..])
+        .success()
+        .stdout_str()
+        .to_string();
+
+    let explicit_double_output = new_ucmd!()
+        .arg("--endian=little")
+        .arg("-An")
+        .arg("-tfD")
+        .run_piped_stdin(&input[..])
+        .success()
+        .stdout_str()
+        .to_string();
+
+    let explicit_float_output = new_ucmd!()
+        .arg("--endian=little")
+        .arg("-An")
+        .arg("-tfF")
+        .run_piped_stdin(&input[..])
+        .success()
+        .stdout_str()
+        .to_string();
+
+    assert_eq!(default_output, explicit_double_output);
+    assert_ne!(default_output, explicit_float_output);
+}
+
+#[test]
+fn test_tf_explicit_float_still_uses_4_bytes() {
+    let input: [u8; 8] = [0x00, 0x00, 0x80, 0x3f, 0x00, 0x00, 0x00, 0x40];
+    new_ucmd!()
+        .arg("--endian=little")
+        .arg("-An")
+        .arg("-tfF")
+        .run_piped_stdin(&input[..])
+        .success()
+        .stdout_only("       1.0000000       2.0000000\n");
+}
+
 #[test]
 fn test_f16() {
     let input: [u8; 14] = [