diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 6753492..b486142 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -30,7 +30,8 @@ jobs:
       - uses: actions-rs/cargo@v1
         with:
           command: build
-          args: --all-features
+          # --all-targets ensures that we also build the benchmarks and tests already.
+          args: --all-features --all-targets
 
       - uses: actions-rs/cargo@v1
         with:
diff --git a/CHANGELOG.md b/CHANGELOG.md
index d3264fd..06233fd 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -7,13 +7,15 @@ adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 ## [Unreleased]
 
 ### Added
-- Added generic support for wrapping mutexes that implement the traits provided by the
+- Generic support for wrapping mutexes that implement the traits provided by the
   [`lock_api`][lock_api] crate. This can be used for creating support for other mutex providers that
   implement it.
 
-- Added support for [`parking_lot`][parking_lot] mutexes. Support includes type aliases for all
+- Support for [`parking_lot`][parking_lot] mutexes. Support includes type aliases for all
   provided mutex types as well as a dedicated `Once` wrapper.
 
+- Simple benchmark to track the rough performance penalty incurred by dependency tracking.
+
 ## [0.1.2] - 2021-05-27
 
 ### Added
diff --git a/Cargo.toml b/Cargo.toml
index d48a6b4..47bcb36 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -17,8 +17,13 @@ lock_api = { version = "0.4", optional = true }
 parking_lot = { version = "0.11", optional = true }
 
 [dev-dependencies]
+criterion = "0.3"
 rand = "0.8"
 
+[[bench]]
+name = "mutex"
+harness = false
+
 [features]
 # Feature names do not match crate names pending namespaced features.
 lockapi = ["lock_api"]
diff --git a/benches/mutex.rs b/benches/mutex.rs
new file mode 100644
index 0000000..0c75198
--- /dev/null
+++ b/benches/mutex.rs
@@ -0,0 +1,82 @@
+use std::sync::Arc;
+use std::sync::Mutex;
+
+use criterion::criterion_group;
+use criterion::criterion_main;
+use criterion::BenchmarkId;
+use criterion::Criterion;
+use criterion::Throughput;
+use rand::prelude::*;
+use tracing_mutex::stdsync::TracingMutex;
+
+const SAMPLE_SIZES: [usize; 5] = [10, 30, 100, 300, 1000];
+
+/// Reproducibly generate random combinations a, b where the index(a) < index(b)
+///
+/// All combinations are generated
+fn generate_combinations<T>(options: &[Arc<T>]) -> Vec<(Arc<T>, Arc<T>)> {
+    let mut combinations = Vec::new();
+
+    for (i, first) in options.iter().enumerate() {
+        for second in options.iter().skip(i + 1) {
+            combinations.push((Arc::clone(first), Arc::clone(second)));
+        }
+    }
+
+    let mut rng = StdRng::seed_from_u64(42);
+
+    combinations.shuffle(&mut rng);
+
+    combinations
+}
+
+/// Take two arbitrary mutexes, lock the first, lock the second while holding the first.
+fn benchmark_baseline(c: &mut Criterion) {
+    let mut group = c.benchmark_group("baseline");
+
+    for nodes in SAMPLE_SIZES {
+        group.throughput(Throughput::Elements((nodes * (nodes - 1) / 2) as u64));
+        group.bench_with_input(BenchmarkId::from_parameter(nodes), &nodes, |b, &s| {
+            b.iter_batched(
+                || {
+                    let mutexes: Vec<_> = (0..s).map(|_| Arc::new(Mutex::new(()))).collect();
+                    generate_combinations(&mutexes)
+                },
+                |combinations| {
+                    for (first, second) in combinations {
+                        let _first = first.lock();
+                        let _second = second.lock();
+                    }
+                },
+                criterion::BatchSize::SmallInput,
+            )
+        });
+    }
+}
+
+/// Same as [`benchmark_baseline`] but now while tracking dependencies.
+fn benchmark_tracing_mutex(c: &mut Criterion) {
+    let mut group = c.benchmark_group("tracing_mutex");
+
+    for nodes in SAMPLE_SIZES {
+        group.throughput(Throughput::Elements((nodes * (nodes - 1) / 2) as u64));
+        group.bench_with_input(BenchmarkId::from_parameter(nodes), &nodes, |b, &s| {
+            b.iter_batched(
+                || {
+                    let mutexes: Vec<_> = (0..s).map(|_| Arc::new(TracingMutex::new(()))).collect();
+                    generate_combinations(&mutexes)
+                },
+                |combinations| {
+                    for (first, second) in combinations {
+                        let _first = first.lock();
+                        let _second = second.lock();
+                    }
+                },
+                criterion::BatchSize::SmallInput,
+            )
+        });
+    }
+}
+
+criterion_group!(benches, benchmark_baseline, benchmark_tracing_mutex);
+criterion_main!(benches);