diff --git a/benchmarks/pandas/bench_dropna.py b/benchmarks/pandas/bench_dropna.py
new file mode 100644
index 00000000..4ccb372e
--- /dev/null
+++ b/benchmarks/pandas/bench_dropna.py
@@ -0,0 +1,43 @@
+"""
+Benchmark: dropna on Series and DataFrame (axis=0, how=any, how=all)
+"""
+import json
+import time
+import numpy as np
+import pandas as pd
+
+ROWS = 100_000
+WARMUP = 3
+ITERATIONS = 10
+
+rng = np.random.default_rng(42)
+series_data = rng.standard_normal(ROWS)
+series_data[::10] = np.nan
+s = pd.Series(series_data)
+
+col_a = rng.standard_normal(ROWS)
+col_b = rng.standard_normal(ROWS)
+col_c = rng.standard_normal(ROWS)
+col_a[::7] = np.nan
+col_b[::11] = np.nan
+col_c[::13] = np.nan
+df = pd.DataFrame({"a": col_a, "b": col_b, "c": col_c})
+
+for _ in range(WARMUP):
+    s.dropna()
+    df.dropna(how="any")
+    df.dropna(how="all")
+
+start = time.perf_counter()
+for _ in range(ITERATIONS):
+    s.dropna()
+    df.dropna(how="any")
+    df.dropna(how="all")
+total = (time.perf_counter() - start) * 1000
+
+print(json.dumps({
+    "function": "dropna",
+    "mean_ms": total / ITERATIONS,
+    "iterations": ITERATIONS,
+    "total_ms": total,
+}))
diff --git a/benchmarks/pandas/bench_fillna.py b/benchmarks/pandas/bench_fillna.py
new file mode 100644
index 00000000..04c60156
--- /dev/null
+++ b/benchmarks/pandas/bench_fillna.py
@@ -0,0 +1,43 @@
+"""
+Benchmark: fillna on Series and DataFrame (scalar, ffill, bfill)
+"""
+import json
+import time
+import numpy as np
+import pandas as pd
+
+ROWS = 100_000
+WARMUP = 3
+ITERATIONS = 10
+
+rng = np.random.default_rng(42)
+series_data = rng.standard_normal(ROWS)
+series_data[::10] = np.nan
+s = pd.Series(series_data)
+
+col_a = rng.standard_normal(ROWS)
+col_b = rng.standard_normal(ROWS)
+col_a[::7] = np.nan
+col_b[::11] = np.nan
+df = pd.DataFrame({"a": col_a, "b": col_b})
+
+for _ in range(WARMUP):
+    s.fillna(0)
+    s.ffill()
+    df.fillna(0)
+    df.bfill()
+
+start = time.perf_counter()
+for _ in range(ITERATIONS):
+    s.fillna(0)
+    s.ffill()
+    df.fillna(0)
+    df.bfill()
+total = (time.perf_counter() - start) * 1000
+
+print(json.dumps({
+    "function": "fillna",
+    "mean_ms": total / ITERATIONS,
+    "iterations": ITERATIONS,
+    "total_ms": total,
+}))
diff --git a/benchmarks/tsb/bench_dropna.ts b/benchmarks/tsb/bench_dropna.ts
new file mode 100644
index 00000000..ecad34e7
--- /dev/null
+++ b/benchmarks/tsb/bench_dropna.ts
@@ -0,0 +1,42 @@
+/**
+ * Benchmark: dropna on Series and DataFrame (axis=0 and axis=1)
+ */
+import { Series, DataFrame, dropna, dropnaDataFrame } from "../../src/index.js";
+
+const ROWS = 100_000;
+const WARMUP = 3;
+const ITERATIONS = 10;
+
+// ~10% missing values
+const seriesData = Float64Array.from({ length: ROWS }, (_, i) =>
+  i % 10 === 0 ? NaN : i * 1.1,
+);
+const s = new Series(seriesData);
+
+const colA = Float64Array.from({ length: ROWS }, (_, i) => (i % 7 === 0 ? NaN : i * 0.5));
+const colB = Float64Array.from({ length: ROWS }, (_, i) => (i % 11 === 0 ? NaN : i * 1.5));
+const colC = Float64Array.from({ length: ROWS }, (_, i) => (i % 13 === 0 ? NaN : i * 2.0));
+const df = DataFrame.fromColumns({ a: colA, b: colB, c: colC });
+
+for (let i = 0; i < WARMUP; i++) {
+  dropna(s);
+  dropnaDataFrame(df, { how: "any" });
+  dropnaDataFrame(df, { how: "all" });
+}
+
+const start = performance.now();
+for (let i = 0; i < ITERATIONS; i++) {
+  dropna(s);
+  dropnaDataFrame(df, { how: "any" });
+  dropnaDataFrame(df, { how: "all" });
+}
+const total = performance.now() - start;
+
+console.log(
+  JSON.stringify({
+    function: "dropna",
+    mean_ms: total / ITERATIONS,
+    iterations: ITERATIONS,
+    total_ms: total,
+  }),
+);
diff --git a/benchmarks/tsb/bench_fillna.ts b/benchmarks/tsb/bench_fillna.ts
new file mode 100644
index 00000000..f56193a7
--- /dev/null
+++ b/benchmarks/tsb/bench_fillna.ts
@@ -0,0 +1,42 @@
+/**
+ * Benchmark: fillna on Series and DataFrame (scalar, ffill, bfill)
+ */
+import { Series, DataFrame, fillnaSeries, fillnaDataFrame } from "../../src/index.js";
+
+const ROWS = 100_000;
+const WARMUP = 3;
+const ITERATIONS = 10;
+
+const seriesData = Float64Array.from({ length: ROWS }, (_, i) =>
+  i % 10 === 0 ? NaN : i * 1.1,
+);
+const s = new Series(seriesData);
+
+const colA = Float64Array.from({ length: ROWS }, (_, i) => (i % 7 === 0 ? NaN : i * 0.5));
+const colB = Float64Array.from({ length: ROWS }, (_, i) => (i % 11 === 0 ? NaN : i * 1.5));
+const df = DataFrame.fromColumns({ a: colA, b: colB });
+
+for (let i = 0; i < WARMUP; i++) {
+  fillnaSeries(s, { value: 0 });
+  fillnaSeries(s, { method: "ffill" });
+  fillnaDataFrame(df, { value: 0 });
+  fillnaDataFrame(df, { method: "bfill" });
+}
+
+const start = performance.now();
+for (let i = 0; i < ITERATIONS; i++) {
+  fillnaSeries(s, { value: 0 });
+  fillnaSeries(s, { method: "ffill" });
+  fillnaDataFrame(df, { value: 0 });
+  fillnaDataFrame(df, { method: "bfill" });
+}
+const total = performance.now() - start;
+
+console.log(
+  JSON.stringify({
+    function: "fillna",
+    mean_ms: total / ITERATIONS,
+    iterations: ITERATIONS,
+    total_ms: total,
+  }),
+);