diff --git a/Cargo.toml b/Cargo.toml
index 11c8c34..11571b7 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -20,7 +20,7 @@ codecov = { repository = "nasso/rs621" }
 [features]
 default = ["rate-limit", "reqwest/default-tls"]
 socks = ["reqwest/socks"]
-rate-limit = ["tokio", "tokio/time", "tokio/sync"]
+rate-limit = ["gloo-timers", "futures", "web-time", "tokio"]
 
 [dependencies]
 thiserror = "1"
@@ -33,8 +33,15 @@ derivative = "2"
 itertools = "0.10"
 futures = { version = "0.3", default-features = false }
 reqwest = { version = ">=0.11, <0.13", default-features = false, features = ["json"] }
-tokio = { optional = true, version = "1" }
 
 [dev-dependencies]
 mockito = "0.30"
 tokio = { version = "1", features = ["rt-multi-thread", "macros"] }
+
+[target.'cfg(target_family = "wasm")'.dependencies]
+gloo-timers = { optional = true, version = "0.3", features = ["futures"] }
+futures = { optional = true, version = "0.3", features = ["std", "alloc"] }
+web-time = { optional = true, version = "1.1.0" }
+
+[target.'cfg(not(target_family = "wasm"))'.dependencies]
+tokio = { optional = true, version = "1", features = ["time", "sync"] }
diff --git a/src/client.rs b/src/client.rs
index 1cfdc07..9bbfd66 100644
--- a/src/client.rs
+++ b/src/client.rs
@@ -1,17 +1,27 @@
-#[cfg(feature = "rate-limit")]
+#[cfg(all(target_family = "wasm", feature = "rate-limit"))]
+#[path = "client/gloo_rate_limit.rs"]
+mod rate_limit;
+
+#[cfg(all(not(target_family = "wasm"), feature = "rate-limit"))]
+#[path = "client/tokio_rate_limit.rs"]
 mod rate_limit;
 
 #[cfg(not(feature = "rate-limit"))]
 #[path = "client/dummy_rate_limit.rs"]
 mod rate_limit;
 
+/// Forced cool down duration performed at every request. E621 allows at most 2 requests per second,
+/// so the lowest safe value we can have here is 500 ms.
+#[cfg(feature = "rate-limit")]
+const REQ_COOLDOWN_DURATION: std::time::Duration = std::time::Duration::from_millis(600);
+
 use futures::Future;
 use reqwest::{Response, Url};
 use serde::Serialize;
 
 use {
     super::error::{Error, Result},
-    reqwest::header::{HeaderMap, HeaderValue},
+    reqwest::header::HeaderMap,
 };
 
 #[cfg(any(target_arch = "wasm32", target_arch = "wasm64"))]
@@ -29,7 +39,7 @@ fn create_header_map<T: AsRef<[u8]>>(user_agent: T) -> Result<HeaderMap> {
         let mut headers = HeaderMap::new();
         headers.insert(
             reqwest::header::USER_AGENT,
-            HeaderValue::from_bytes(user_agent.as_ref())
+            reqwest::header::HeaderValue::from_bytes(user_agent.as_ref())
                 .map_err(|e| Error::InvalidHeaderValue(format!("{}", e)))?,
         );
 
diff --git a/src/client/gloo_rate_limit.rs b/src/client/gloo_rate_limit.rs
new file mode 100644
index 0000000..5b1eead
--- /dev/null
+++ b/src/client/gloo_rate_limit.rs
@@ -0,0 +1,53 @@
+use super::REQ_COOLDOWN_DURATION;
+
+use futures::lock::{Mutex, MutexGuard};
+
+use std::future::Future;
+use std::sync::Arc;
+
+use web_time::Instant;
+
+#[derive(Debug, Clone, Default)]
+pub struct RateLimit {
+    // Use a `futures` `Mutex` because ~500ms is crazy long to block an async task.
+    deadline: Arc<Mutex<Option<Instant>>>,
+}
+
+struct Guard<'a>(MutexGuard<'a, Option<Instant>>);
+
+impl<'a> Drop for Guard<'a> {
+    fn drop(&mut self) {
+        // Use a `Drop` impl so that updating the deadline is panic-safe.
+        *self.0 = Some(Instant::now() + REQ_COOLDOWN_DURATION);
+    }
+}
+
+impl RateLimit {
+    async fn lock(&self) -> Guard {
+        loop {
+            let now = Instant::now();
+
+            let deadline = {
+                let guard = self.deadline.lock().await;
+
+                match &*guard {
+                    None => return Guard(guard),
+                    Some(deadline) if now >= *deadline => return Guard(guard),
+                    Some(deadline) => *deadline,
+                }
+            };
+
+            gloo_timers::future::sleep(deadline - now).await;
+        }
+    }
+
+    pub async fn check<F, R>(self, fut: F) -> R
+    where
+        F: Future<Output = R>,
+    {
+        let guard = self.lock().await;
+        let result = fut.await;
+        drop(guard);
+        result
+    }
+}
diff --git a/src/client/rate_limit.rs b/src/client/tokio_rate_limit.rs
similarity index 80%
rename from src/client/rate_limit.rs
rename to src/client/tokio_rate_limit.rs
index f3bfc64..1a63a70 100644
--- a/src/client/rate_limit.rs
+++ b/src/client/tokio_rate_limit.rs
@@ -1,13 +1,11 @@
-use futures::Future;
+use super::REQ_COOLDOWN_DURATION;
+
+use std::future::Future;
 
 use std::sync::Arc;
 
 use tokio::sync::{Mutex, MutexGuard};
-use tokio::time::{sleep_until, Duration, Instant};
-
-/// Forced cool down duration performed at every request. E621 allows at most 2 requests per second,
-/// so the lowest safe value we can have here is 500 ms.
-const REQ_COOLDOWN_DURATION: Duration = Duration::from_millis(600);
+use tokio::time::{sleep_until, Instant};
 
 #[derive(Debug, Clone, Default)]
 pub struct RateLimit {