diff --git a/docs/rate-limiting.md b/docs/rate-limiting.md
index 6ab1498..9e3ae66 100644
--- a/docs/rate-limiting.md
+++ b/docs/rate-limiting.md
@@ -9,6 +9,11 @@ window (default 5 requests per minute) to reduce abuse from repeated challenge o
 verification attempts. The limiter uses the submitted Stellar address when present
 and falls back to the client IP otherwise.
 
+User-facing routes under `/api/users` (excluding `/api/users/health`) use a
+per-user fixed-window limiter (default **60 requests per minute**). Authenticated
+callers are keyed by database user id (`users:{id}`); anonymous callers fall back
+to IP. See [`docs/users-rate-limiting.md`](./users-rate-limiting.md).
+
 ## Configuration
 
 | Variable | Default | Description |
diff --git a/docs/users-rate-limiting.md b/docs/users-rate-limiting.md
new file mode 100644
index 0000000..ad601e6
--- /dev/null
+++ b/docs/users-rate-limiting.md
@@ -0,0 +1,54 @@
+# Users API rate limiting
+
+Authenticated and anonymous requests under `GET /api/users/*` (via `usersRouter`)
+use a per-identity fixed-window rate limit powered by
+[`createPerUserRateLimiter`](../src/middleware/rateLimit.ts) (IETF draft-7 headers).
+
+## Policy
+
+| Setting | Value |
+|---------|--------|
+| Window | 60 seconds |
+| Limit | 60 requests per key |
+| Authenticated key | `users:{user.id}` (after `requireAuthForbidden` on `/me`) |
+| Anonymous key | `users:ip:{socket.remoteAddress}` (public profile / predictions) |
+
+`GET /api/users/me` runs auth **before** the limiter so the bucket is per
+database user id. Public GETs are IP-keyed (Bearer on those paths does not
+change the key unless `req.user` is already populated by upstream middleware).
+
+`/api/users/health` is mounted on a separate router and is **not** rate-limited.
+
+Social (`socialRouter`) and portfolio (`userPortfolioRouter`) mounts under
+`/api/users` are unchanged by this policy.
+
+## 429 response
+
+```json
+{
+  "error": {
+    "code": "rate_limit_exceeded",
+    "message": "Too many requests",
+    "retryAfter": 12,
+    "resetAt": "2026-07-24T12:00:00.000Z"
+  }
+}
+```
+
+Also includes:
+
+- `Retry-After` header (seconds)
+- Draft-7 `RateLimit-Limit` / `RateLimit-Remaining` / `RateLimit-Reset` headers
+- Audit event `rate_limit.blocked` with correlation ID
+
+## Implementation
+
+- Router: [`src/routes/users.ts`](../src/routes/users.ts)
+- Middleware: [`src/middleware/rateLimit.ts`](../src/middleware/rateLimit.ts)
+- Tests: [`tests/usersRateLimit.test.ts`](../tests/usersRateLimit.test.ts)
+
+## Verification
+
+```bash
+npm test -- tests/usersRateLimit.test.ts
+```
diff --git a/src/middleware/rateLimit.ts b/src/middleware/rateLimit.ts
index d28728b..d54ca13 100644
--- a/src/middleware/rateLimit.ts
+++ b/src/middleware/rateLimit.ts
@@ -148,6 +148,16 @@ export function createRateLimiter(options: Partial<Options> = {}): RateLimitRequ
   }) as RateLimitRequestHandler;
 }
 
+/**
+ * Fixed-window per-identity limiter (default 60 requests / 60 seconds).
+ *
+ * Prefer an explicit `keyGenerator` when mounting on a route family so buckets
+ * stay isolated (e.g. `users:{id}` on `/api/users`, `predictions:{id}` on
+ * `/api/predictions`). Falls back to `user:{address|sub|id}` then `ip:{ip}`.
+ *
+ * Emits IETF draft-7 `RateLimit-*` headers and the standard
+ * `{ error: { code: "rate_limit_exceeded", ... } }` envelope on 429.
+ */
 export function createPerUserRateLimiter(options: Partial<Options> = {}): RateLimitRequestHandler {
   return createRateLimiter({
     windowMs: 60 * 1000,
diff --git a/src/routes/users.ts b/src/routes/users.ts
index 63336ea..c1ada1b 100644
--- a/src/routes/users.ts
+++ b/src/routes/users.ts
@@ -21,6 +21,12 @@
  *     → req.id → new UUID) and stamps it on `res.locals.correlationId`.
  *   - Echoes the correlation ID back via the X-Correlation-Id response header.
  *   - Emits a structured `users_access_log` entry on every response finish.
+ *
+ * Rate limiting (issue #411 / users-rl-v7):
+ *   - `createPerUserRateLimiter` (60 req/min, IETF draft-7 headers) on each route.
+ *   - `GET /me` authenticates first, then keys by `users:{user.id}`.
+ *   - Public GETs fall back to `users:ip:{ip}` (no soft auth — preserves 403/401 contracts).
+ *   - `/api/users/health` is mounted separately and is not throttled here.
  */
 
 import { Router, Request, Response, NextFunction } from "express";
@@ -34,6 +40,7 @@ import {
 import { requireAuthForbidden } from "../middleware/requireAuth";
 import { AuthenticatedRequest } from "../middleware/auth";
 import { accessLog } from "../middleware/accessLog";
+import { createPerUserRateLimiter } from "../middleware/rateLimit";
 import { logger } from "../config/logger";
 import { getRequestId } from "../lib/requestContext";
 import { clampLimit, DEFAULT_PAGE_SIZE } from "../utils/cursor";
@@ -48,6 +55,22 @@ const stellarAddressSchema = z
   .string()
   .regex(/^G[A-Z2-7]{55}$/, "Invalid Stellar address");
 
+/**
+ * Shared /api/users limiter. Authenticated requests (req.user set) use
+ * `users:{id}`; anonymous traffic uses `users:ip:{ip}`.
+ */
+const usersRateLimit = createPerUserRateLimiter({
+  windowMs: 60 * 1000,
+  limit: 60,
+  keyGenerator: (req) => {
+    const userId = (req as AuthenticatedRequest).user?.id;
+    if (typeof userId === "string" && userId.trim().length > 0) {
+      return `users:${userId}`;
+    }
+    return `users:ip:${req.socket?.remoteAddress ?? "unknown"}`;
+  },
+});
+
 // ---------------------------------------------------------------------------
 // Access log — must be the first middleware so every handler inherits the
 // correlation ID via res.locals.correlationId.
@@ -67,7 +90,11 @@ usersRouter.use(usersMetricsMiddleware);
 // ---------------------------------------------------------------------------
 // GET /api/users/me
 // ---------------------------------------------------------------------------
-usersRouter.get("/me", requireAuthForbidden, async (req: AuthenticatedRequest, res, next) => {
+usersRouter.get(
+  "/me",
+  requireAuthForbidden,
+  usersRateLimit,
+  async (req: AuthenticatedRequest, res, next) => {
   const correlationId = res.locals.correlationId as string;
 
   try {
@@ -121,6 +148,7 @@ usersRouter.get("/me", requireAuthForbidden, async (req: AuthenticatedRequest, r
  */
 usersRouter.get(
   "/:address/predictions",
+  usersRateLimit,
   async (req: Request, res: Response, next: NextFunction) => {
     // Prefer the access-log correlation ID; fall back to ALS for non-route callers.
     const correlationId = (res.locals.correlationId as string | undefined) ?? getRequestId();
@@ -201,6 +229,7 @@ usersRouter.get(
 // ---------------------------------------------------------------------------
 usersRouter.get(
   "/:stellarAddress/profile",
+  usersRateLimit,
   async (req, res, next) => {
     const correlationId = (res.locals.correlationId as string | undefined) ?? getRequestId();
     const reqId = correlationId;
diff --git a/tests/usersRateLimit.test.ts b/tests/usersRateLimit.test.ts
new file mode 100644
index 0000000..2a26a45
--- /dev/null
+++ b/tests/usersRateLimit.test.ts
@@ -0,0 +1,130 @@
+/**
+ * Rate-limit coverage for /api/users (issue #411 / users-rl-v7).
+ *
+ * Mirrors the production mount in `src/routes/users.ts`:
+ *   - key `users:{user.id}` when authenticated
+ *   - key `users:ip:{ip}` otherwise
+ *   - standard 429 envelope + draft-7 RateLimit headers
+ */
+
+import request from "supertest";
+import express from "express";
+import type { Request } from "express";
+import { createPerUserRateLimiter } from "../src/middleware/rateLimit";
+
+jest.mock("../src/services/auditService", () => ({
+  createAuditLog: jest.fn().mockResolvedValue(undefined),
+}));
+
+type TestUser = { id?: string; address?: string };
+
+/** Same keying rules as `usersRouter` (low limit for fast tests). */
+function usersRateLimiter(limit = 2) {
+  return createPerUserRateLimiter({
+    windowMs: 60_000,
+    limit,
+    keyGenerator: (req: Request) => {
+      const userId = (req as Request & { user?: TestUser }).user?.id;
+      if (typeof userId === "string" && userId.trim().length > 0) {
+        return `users:${userId}`;
+      }
+      return `users:ip:${req.socket?.remoteAddress ?? "unknown"}`;
+    },
+  });
+}
+
+function makeApp(limit = 2) {
+  const app = express();
+  app.use((req, _res, next) => {
+    const userId = req.headers["x-test-user-id"];
+    if (typeof userId === "string" && userId.trim().length > 0) {
+      (req as Request & { user?: TestUser }).user = { id: userId };
+    }
+    next();
+  });
+  app.use(usersRateLimiter(limit));
+  app.get("/api/users/me", (_req, res) => {
+    res.json({ data: { ok: true } });
+  });
+  app.get("/api/users/:address/profile", (_req, res) => {
+    res.json({ data: { ok: true } });
+  });
+  return app;
+}
+
+describe("per-user rate limiting for /api/users", () => {
+  it("enforces the limit independently for each authenticated user id", async () => {
+    const app = makeApp();
+
+    expect((await request(app).get("/api/users/me").set("x-test-user-id", "user-a")).status).toBe(
+      200,
+    );
+    expect((await request(app).get("/api/users/me").set("x-test-user-id", "user-a")).status).toBe(
+      200,
+    );
+    expect((await request(app).get("/api/users/me").set("x-test-user-id", "user-a")).status).toBe(
+      429,
+    );
+
+    const otherUser = await request(app).get("/api/users/me").set("x-test-user-id", "user-b");
+    expect(otherUser.status).toBe(200);
+  });
+
+  it("falls back to a per-IP bucket when no user id is present", async () => {
+    const app = makeApp();
+
+    expect((await request(app).get("/api/users/GTEST/profile")).status).toBe(200);
+    expect((await request(app).get("/api/users/GTEST/profile")).status).toBe(200);
+    expect((await request(app).get("/api/users/GTEST/profile")).status).toBe(429);
+  });
+
+  it("does not share the IP bucket with an authenticated user bucket", async () => {
+    const app = makeApp();
+
+    // Exhaust anonymous IP bucket
+    await request(app).get("/api/users/GTEST/profile");
+    await request(app).get("/api/users/GTEST/profile");
+    expect((await request(app).get("/api/users/GTEST/profile")).status).toBe(429);
+
+    // Authenticated identity still has its own quota
+    const authenticated = await request(app)
+      .get("/api/users/me")
+      .set("x-test-user-id", "user-isolated");
+    expect(authenticated.status).toBe(200);
+  });
+
+  it("returns the standard rate-limit error envelope", async () => {
+    const app = makeApp();
+
+    await request(app).get("/api/users/me").set("x-test-user-id", "user-envelope");
+    await request(app).get("/api/users/me").set("x-test-user-id", "user-envelope");
+    const response = await request(app)
+      .get("/api/users/me")
+      .set("x-test-user-id", "user-envelope");
+
+    expect(response.status).toBe(429);
+    expect(response.body.error).toMatchObject({
+      code: "rate_limit_exceeded",
+      message: "Too many requests",
+    });
+    expect(Number(response.headers["retry-after"])).toBeGreaterThanOrEqual(1);
+    expect(response.body.error.retryAfter).toBe(Number(response.headers["retry-after"]));
+    expect(typeof response.body.error.resetAt).toBe("string");
+  });
+
+  it("exposes IETF draft-7 RateLimit headers on successful responses", async () => {
+    const app = makeApp(3);
+
+    const response = await request(app)
+      .get("/api/users/me")
+      .set("x-test-user-id", "user-headers");
+
+    expect(response.status).toBe(200);
+    // express-rate-limit `standardHeaders: "draft-7"` emits a combined RateLimit header.
+    const combined =
+      response.headers["ratelimit"] ?? response.headers["RateLimit"];
+    expect(combined).toBeDefined();
+    expect(String(combined)).toMatch(/limit=\d+/i);
+    expect(String(combined)).toMatch(/remaining=\d+/i);
+  });
+});