diff --git a/xxhash.go b/xxhash.go
index a9e0d45..78bddf1 100644
--- a/xxhash.go
+++ b/xxhash.go
@@ -19,10 +19,13 @@ const (
 // Store the primes in an array as well.
 //
 // The consts are used when possible in Go code to avoid MOVs but we need a
-// contiguous array of the assembly code.
+// contiguous array for the assembly code.
 var primes = [...]uint64{prime1, prime2, prime3, prime4, prime5}
 
 // Digest implements hash.Hash64.
+//
+// Note that a zero-valued Digest is not ready to receive writes.
+// Call Reset or create a Digest using New before calling other methods.
 type Digest struct {
 	v1    uint64
 	v2    uint64
@@ -33,19 +36,31 @@ type Digest struct {
 	n     int // how much of mem is used
 }
 
-// New creates a new Digest that computes the 64-bit xxHash algorithm.
+// New creates a new Digest with a zero seed.
 func New() *Digest {
+	return NewWithSeed(0)
+}
+
+// NewWithSeed creates a new Digest with the given seed.
+func NewWithSeed(seed uint64) *Digest {
 	var d Digest
-	d.Reset()
+	d.ResetWithSeed(seed)
 	return &d
 }
 
 // Reset clears the Digest's state so that it can be reused.
+// It uses a seed value of zero.
 func (d *Digest) Reset() {
-	d.v1 = primes[0] + prime2
-	d.v2 = prime2
-	d.v3 = 0
-	d.v4 = -primes[0]
+	d.ResetWithSeed(0)
+}
+
+// ResetWithSeed clears the Digest's state so that it can be reused.
+// It uses the given seed to initialize the state.
+func (d *Digest) ResetWithSeed(seed uint64) {
+	d.v1 = seed + prime1 + prime2
+	d.v2 = seed + prime2
+	d.v3 = seed
+	d.v4 = seed - prime1
 	d.total = 0
 	d.n = 0
 }
diff --git a/xxhash_asm.go b/xxhash_asm.go
index 9216e0a..78f95f2 100644
--- a/xxhash_asm.go
+++ b/xxhash_asm.go
@@ -6,7 +6,7 @@
 
 package xxhash
 
-// Sum64 computes the 64-bit xxHash digest of b.
+// Sum64 computes the 64-bit xxHash digest of b with a zero seed.
 //
 //go:noescape
 func Sum64(b []byte) uint64
diff --git a/xxhash_other.go b/xxhash_other.go
index 26df13b..118e49e 100644
--- a/xxhash_other.go
+++ b/xxhash_other.go
@@ -3,7 +3,7 @@
 
 package xxhash
 
-// Sum64 computes the 64-bit xxHash digest of b.
+// Sum64 computes the 64-bit xxHash digest of b with a zero seed.
 func Sum64(b []byte) uint64 {
 	// A simpler version would be
 	//   d := New()
diff --git a/xxhash_safe.go b/xxhash_safe.go
index e86f1b5..05f5e7d 100644
--- a/xxhash_safe.go
+++ b/xxhash_safe.go
@@ -5,7 +5,7 @@
 
 package xxhash
 
-// Sum64String computes the 64-bit xxHash digest of s.
+// Sum64String computes the 64-bit xxHash digest of s with a zero seed.
 func Sum64String(s string) uint64 {
 	return Sum64([]byte(s))
 }
diff --git a/xxhash_test.go b/xxhash_test.go
index 6330f19..8e2f456 100644
--- a/xxhash_test.go
+++ b/xxhash_test.go
@@ -4,45 +4,60 @@ import (
 	"bytes"
 	"encoding/binary"
 	"fmt"
+	"math"
 	"strings"
 	"testing"
 )
 
 func TestAll(t *testing.T) {
+	// Exactly 63 characters, which exercises all code paths.
+	const s63 = "Call me Ishmael. Some years ago--never mind how long precisely-"
 	for _, tt := range []struct {
-		name  string
 		input string
+		seed  uint64
 		want  uint64
 	}{
-		{"empty", "", 0xef46db3751d8e999},
-		{"a", "a", 0xd24ec4f1a98c6e5b},
-		{"as", "as", 0x1c330fb2d66be179},
-		{"asd", "asd", 0x631c37ce72a97393},
-		{"asdf", "asdf", 0x415872f599cea71e},
-		{
-			"len=63",
-			// Exactly 63 characters, which exercises all code paths.
-			"Call me Ishmael. Some years ago--never mind how long precisely-",
-			0x02a2e85470d6fd96,
-		},
+		{"", 0, 0xef46db3751d8e999},
+		{"a", 0, 0xd24ec4f1a98c6e5b},
+		{"as", 0, 0x1c330fb2d66be179},
+		{"asd", 0, 0x631c37ce72a97393},
+		{"asdf", 0, 0x415872f599cea71e},
+		{s63, 0, 0x02a2e85470d6fd96},
+
+		{"", 123, 0xe0db84de91f3e198},
+		{"asdf", math.MaxUint64, 0x9a2fd8473be539b6},
+		{s63, 54321, 0x1736d186daf5d1cd},
 	} {
 		lastChunkSize := len(tt.input)
 		if lastChunkSize == 0 {
 			lastChunkSize = 1
 		}
+		var name string
+		if tt.input == "" {
+			name = "input=empty"
+		} else if len(tt.input) > 10 {
+			name = fmt.Sprintf("input=len-%d", len(tt.input))
+		} else {
+			name = fmt.Sprintf("input=%q", tt.input)
+		}
+		if tt.seed != 0 {
+			name += fmt.Sprintf(",seed=%d", tt.seed)
+		}
 		for chunkSize := 1; chunkSize <= lastChunkSize; chunkSize++ {
-			name := fmt.Sprintf("%s,chunkSize=%d", tt.name, chunkSize)
+			name := fmt.Sprintf("%s,chunkSize=%d", name, chunkSize)
 			t.Run(name, func(t *testing.T) {
-				testDigest(t, tt.input, chunkSize, tt.want)
+				testDigest(t, tt.input, tt.seed, chunkSize, tt.want)
 			})
 		}
-		t.Run(tt.name, func(t *testing.T) { testSum(t, tt.input, tt.want) })
+		if tt.seed == 0 {
+			t.Run(name, func(t *testing.T) { testSum(t, tt.input, tt.want) })
+		}
 	}
 }
 
-func testDigest(t *testing.T, input string, chunkSize int, want uint64) {
-	d := New()
-	ds := New() // uses WriteString
+func testDigest(t *testing.T, input string, seed uint64, chunkSize int, want uint64) {
+	d := NewWithSeed(seed)
+	ds := NewWithSeed(seed) // uses WriteString
 	for i := 0; i < len(input); i += chunkSize {
 		chunk := input[i:]
 		if len(chunk) > chunkSize {
@@ -96,6 +111,23 @@ func TestReset(t *testing.T) {
 	}
 }
 
+func TestResetWithSeed(t *testing.T) {
+	parts := []string{"The quic", "k br", "o", "wn fox jumps", " ov", "er the lazy ", "dog."}
+	d := NewWithSeed(123)
+	for _, part := range parts {
+		d.Write([]byte(part))
+	}
+	h0 := d.Sum64()
+
+	d.ResetWithSeed(123)
+	d.Write([]byte(strings.Join(parts, "")))
+	h1 := d.Sum64()
+
+	if h0 != h1 {
+		t.Errorf("0x%x != 0x%x", h0, h1)
+	}
+}
+
 func TestBinaryMarshaling(t *testing.T) {
 	d := New()
 	d.WriteString("abc")
diff --git a/xxhash_unsafe.go b/xxhash_unsafe.go
index 1c1638f..cf9d42a 100644
--- a/xxhash_unsafe.go
+++ b/xxhash_unsafe.go
@@ -33,7 +33,7 @@ import (
 //
 // See https://github.com/golang/go/issues/42739 for discussion.
 
-// Sum64String computes the 64-bit xxHash digest of s.
+// Sum64String computes the 64-bit xxHash digest of s with a zero seed.
 // It may be faster than Sum64([]byte(s)) by avoiding a copy.
 func Sum64String(s string) uint64 {
 	b := *(*[]byte)(unsafe.Pointer(&sliceHeader{s, len(s)}))