diff --git a/README.md b/README.md
index 7708c42..023c23a 100644
--- a/README.md
+++ b/README.md
@@ -150,6 +150,11 @@ from `src/middleware/rateLimiter.ts` unless overridden by configuration. When
 `API_KEY` authentication is configured, the presented key identifies the
 client; open deployments continue to use the client IP.
 
+> **Operational Note (Multi-Instance Deployments):**
+> Rate limiter state lives in a plain `Map` local to the middleware instance. In multi-instance deployments without sticky sessions, clients may receive N× their intended budget (where N is the number of replicas), and limits reset entirely upon instance restart.
+>
+> A shared distributed store (like Redis) is deliberately deferred until a broader persistence layer is introduced to the service, to avoid bloating operational requirements prematurely. However, memory growth is strictly bounded: the internal `Map` is capped at 5000 entries. When capacity is reached, it lazily prunes expired buckets before evicting the oldest entry to protect against memory-pressure attacks.
+
 `POST /api/v1/quote` is excluded from the global limiter via `skipPaths` and
 then receives its own stricter `rateLimiter({ max: 10, windowMs: 60_000 })`
 instance in `src/app.ts`. That quote limiter has separate in-memory counters
diff --git a/src/middleware/rateLimiter.test.ts b/src/middleware/rateLimiter.test.ts
index 7199f68..f969d3d 100644
--- a/src/middleware/rateLimiter.test.ts
+++ b/src/middleware/rateLimiter.test.ts
@@ -182,4 +182,50 @@ describe("rateLimiter", () => {
     const blocked = await request(app).post("/api/v1/quote-history");
     expect(blocked.status).toBe(429);
   });
+
+  it("allows bypass across multiple middleware instances", async () => {
+    const app = express();
+    app.set("trust proxy", true);
+    
+    const limiter1 = rateLimiter({ max: 1, windowMs: 1000 });
+    app.post("/route1", limiter1, (_req, res) => res.status(201).json({ ok: true }));
+    
+    const limiter2 = rateLimiter({ max: 1, windowMs: 1000 });
+    app.post("/route2", limiter2, (_req, res) => res.status(201).json({ ok: true }));
+    
+    app.use(errorHandler);
+
+    // Client hits route1, consumes quota
+    await request(app).post("/route1").set("x-forwarded-for", "10.0.0.1").expect(201);
+    await request(app).post("/route1").set("x-forwarded-for", "10.0.0.1").expect(429);
+
+    // Same client hits route2, gets full quota again
+    await request(app).post("/route2").set("x-forwarded-for", "10.0.0.1").expect(201);
+  });
+
+  it("bounds memory growth by evicting the oldest bucket when capacity is reached", () => {
+    const limiter = rateLimiter({ max: 1, windowMs: 60000 });
+    const next = jest.fn();
+    const res = {} as Response;
+    
+    const req0 = { method: "POST", ip: "client-0", path: "/mutate" } as unknown as Request;
+    limiter(req0, res, next);
+    
+    limiter(req0, res, next);
+    expect(next).toHaveBeenLastCalledWith(expect.objectContaining({ status: 429 }));
+    
+    // Fill the map up to 5000 (MAX_BUCKETS)
+    for (let i = 1; i <= 5000; i++) {
+      const req = { method: "POST", ip: `client-${i}`, path: "/mutate" } as unknown as Request;
+      limiter(req, res, next);
+    }
+    
+    // Because Client 0 was inserted first, inserting client-5000 triggered eviction of client-0.
+    // Client 0 should now be granted a new quota.
+    next.mockClear();
+    limiter(req0, res, next);
+    expect(next).toHaveBeenCalledTimes(1);
+    expect(next).not.toHaveBeenCalledWith(expect.objectContaining({ status: 429 }));
+  });
 });
+
diff --git a/src/middleware/rateLimiter.ts b/src/middleware/rateLimiter.ts
index 6b25d14..cad3b8e 100644
--- a/src/middleware/rateLimiter.ts
+++ b/src/middleware/rateLimiter.ts
@@ -20,6 +20,9 @@ const DEFAULT_MAX = 30;
 /** Default rolling window length, in milliseconds. */
 const DEFAULT_WINDOW_MS = 60_000;
 
+/** Maximum number of buckets tracked in memory to prevent unbounded growth. */
+const MAX_BUCKETS = 5000;
+
 interface Bucket {
   count: number;
   resetAt: number;
@@ -77,6 +80,17 @@ export function rateLimiter(
     const bucket = buckets.get(key);
 
     if (!bucket || bucket.resetAt <= now) {
+      if (!bucket && buckets.size >= MAX_BUCKETS) {
+        for (const [k, v] of buckets.entries()) {
+          if (v.resetAt <= now) buckets.delete(k);
+        }
+        if (buckets.size >= MAX_BUCKETS) {
+          const oldestKey = buckets.keys().next().value;
+          if (oldestKey !== undefined) {
+            buckets.delete(oldestKey);
+          }
+        }
+      }
       buckets.set(key, { count: 1, resetAt: now + windowMs });
       next();
       return;