diff --git a/README.md b/README.md
index 78ac862..8ba154a 100644
--- a/README.md
+++ b/README.md
@@ -34,6 +34,8 @@ The `ec-slimloader` crate will handle for you:
 * it will read from the state journal what image slot will be booted.
 * on subsequent reboots, it will fall back to your defined backup slot if you do not mark your current application image as `confirmed`.
 
+If that backup fails validation and differs from the target, the loader also tries the target before giving up. It records `Attempting` before this reverse fallback, so the application identifies the correct running slot and inactive update slot. Each distinct candidate is checked at most once per boot.
+
 However, some aspects are handled by the platform support crate (and can differ from project-to-project):
 * how application images are loaded. For `ec-slimloader-imxrt` images are copied to RAM in a quite chip-specific way. Typically for other platforms you might want to swap images between on-chip NOR flash and external NOR flash. The latter method is not implemented in this repository (yet).
 * how application images are verified. By default the images themselves are not checked at all. `ec-slimloader-imxrt` leverages the native NXP authentication routines to check image integrity.
diff --git a/libs/ec-slimloader/src/lib.rs b/libs/ec-slimloader/src/lib.rs
index 14cac4c..32de9c9 100644
--- a/libs/ec-slimloader/src/lib.rs
+++ b/libs/ec-slimloader/src/lib.rs
@@ -164,6 +164,16 @@ pub async fn start<B: Board, const JOURNAL_BUFFER_SIZE: usize>(config: B::Config
         info!("Attempting to boot backup in {:?}", slot);
         let error = board.check_and_boot(&state.backup()).await; // If this function returns, it implies that the boot has failed.
         warn!("Failed to boot backup in {:?} because {:?}", slot, error);
+    } else if slot != state.target() {
+        // We started in [Attempting] or [Failed], so the backup was tried first and could not boot.
+        // The target has not been tried this boot; an interrupted earlier attempt may have left it usable.
+        // Try the target as a last resort, recording [Attempting] in the journal so the app
+        // identifies the target as running and writes future updates to the backup.
+
+        set_status::<_, JOURNAL_BUFFER_SIZE>(&mut board, &mut state, Status::Attempting).await;
+        info!("Backup unusable, retrying target {:?}", state.target());
+        let error = board.check_and_boot(&state.target()).await;
+        warn!("Failed to boot target {:?} because {:?}", state.target(), error);
     }
 
     error!("No candidates booted successfully, giving up...");