diff --git a/backend/src/routes/adminEmail.js b/backend/src/routes/adminEmail.js
index acc37156..d880b884 100644
--- a/backend/src/routes/adminEmail.js
+++ b/backend/src/routes/adminEmail.js
@@ -94,6 +94,16 @@ router.post('/config', [
       await db('email_configs').insert(configData);
     }
 
+    // Refresh the cached transporter so the new SMTP settings take effect
+    // immediately. Without this, a previously-initialised transporter stays
+    // cached (the queue processor only re-inits when it's null), so changing
+    // the email account had no effect until a backend restart — emails kept
+    // failing against the old/empty config. initializeTransporter catches its
+    // own errors and returns null, so this never throws; an invalid config
+    // simply leaves the transporter null (surfaced via the Test-email button).
+    const { initializeTransporter } = require('../services/emailProcessor');
+    await initializeTransporter(true);
+
     // Log activity
     await logActivity('email_config_updated', 
       { smtp_host, from_email }, 
diff --git a/backend/src/services/emailProcessor.js b/backend/src/services/emailProcessor.js
index 6b6308d3..0e9b2412 100644
--- a/backend/src/services/emailProcessor.js
+++ b/backend/src/services/emailProcessor.js
@@ -774,13 +774,18 @@ async function processEmailQueue({ ignoreSchedule = false, limit = 10 } = {}) {
       // queue split-payment emails relative to the event date.
       const now = new Date();
       const query = db('email_queue')
-        .where('status', 'pending')
-        .where('retry_count', '<', 3);
+        .where('status', 'pending');
       if (!ignoreSchedule) {
-        query.andWhere(function() {
+        // Automatic runs: respect the retry cap (don't hammer a failing
+        // address) AND the schedule (business-hours floor / future send).
+        query.where('retry_count', '<', 3).andWhere(function() {
           this.whereNull('scheduled_at').orWhere('scheduled_at', '<=', now);
         });
       }
+      // A manual "send now" (ignoreSchedule) deliberately bypasses BOTH the
+      // schedule and the retry cap: the admin is forcing a retry, typically
+      // right after fixing SMTP. Without this, emails that failed 3× during
+      // an SMTP outage are stuck "pending" forever with no way to resend.
       pendingEmails = await query
         .orderBy('scheduled_at', 'asc')
         .orderBy('created_at', 'asc')