diff --git a/doc/site/modules/core/string.markdown b/doc/site/modules/core/string.markdown
index eed1410c..3b1aad57 100644
--- a/doc/site/modules/core/string.markdown
+++ b/doc/site/modules/core/string.markdown
@@ -120,6 +120,15 @@ Returns the index of the first byte matching `search` in the string or `-1` if
 
 It is a runtime error if `search` is not a string.
 
+### **indexOf**(search, start)
+
+Returns the index of the first byte matching `search` in the string or `-1` if
+`search` was not found, starting a byte offset `start`. The start can be
+negative to count backwards from the end of the string.
+
+It is a runtime error if `search` is not a string or `start` is not an integer
+index within the string's byte length.
+
 ### **iterate**(iterator), **iteratorValue**(iterator)
 
 Implements the [iterator protocol](../../control-flow.html#the-iterator-protocol)
diff --git a/src/vm/wren_core.c b/src/vm/wren_core.c
index 51c1e9ab..4f91d3ca 100644
--- a/src/vm/wren_core.c
+++ b/src/vm/wren_core.c
@@ -879,14 +879,14 @@ DEF_PRIMITIVE(string_endsWith)
   ObjString* string = AS_STRING(args[0]);
   ObjString* search = AS_STRING(args[1]);
 
-  // Corner case, if the search string is longer than return false right away.
+  // Edge case: If the search string is longer then return false right away.
   if (search->length > string->length) RETURN_FALSE;
 
   RETURN_BOOL(memcmp(string->value + string->length - search->length,
                      search->value, search->length) == 0);
 }
 
-DEF_PRIMITIVE(string_indexOf)
+DEF_PRIMITIVE(string_indexOf1)
 {
   if (!validateString(vm, args[1], "Argument")) return false;
 
@@ -897,15 +897,16 @@ DEF_PRIMITIVE(string_indexOf)
   RETURN_NUM(index == UINT32_MAX ? -1 : (int)index);
 }
 
-DEF_PRIMITIVE(string_indexOf_with_startIndex)
+DEF_PRIMITIVE(string_indexOf2)
 {
   if (!validateString(vm, args[1], "Argument")) return false;
 
   ObjString* string = AS_STRING(args[0]);
   ObjString* search = AS_STRING(args[1]);
-  uint32_t startIndex = AS_NUM(args[2]);
-
-  uint32_t index = wrenStringFind(string, search, startIndex);
+  uint32_t start = validateIndex(vm, args[2], string->length, "Start");
+  if (start == UINT32_MAX) return false;
+  
+  uint32_t index = wrenStringFind(string, search, start);
   RETURN_NUM(index == UINT32_MAX ? -1 : (int)index);
 }
 
@@ -974,7 +975,7 @@ DEF_PRIMITIVE(string_startsWith)
   ObjString* string = AS_STRING(args[0]);
   ObjString* search = AS_STRING(args[1]);
 
-  // Corner case, if the search string is longer than return false right away.
+  // Edge case: If the search string is longer then return false right away.
   if (search->length > string->length) RETURN_FALSE;
 
   RETURN_BOOL(memcmp(string->value, search->value, search->length) == 0);
@@ -1263,8 +1264,8 @@ void wrenInitializeCore(WrenVM* vm)
   PRIMITIVE(vm->stringClass, "codePointAt_(_)", string_codePointAt);
   PRIMITIVE(vm->stringClass, "contains(_)", string_contains);
   PRIMITIVE(vm->stringClass, "endsWith(_)", string_endsWith);
-  PRIMITIVE(vm->stringClass, "indexOf(_)", string_indexOf);
-  PRIMITIVE(vm->stringClass, "indexOf(_,_)", string_indexOf_with_startIndex);
+  PRIMITIVE(vm->stringClass, "indexOf(_)", string_indexOf1);
+  PRIMITIVE(vm->stringClass, "indexOf(_,_)", string_indexOf2);
   PRIMITIVE(vm->stringClass, "iterate(_)", string_iterate);
   PRIMITIVE(vm->stringClass, "iterateByte_(_)", string_iterateByte);
   PRIMITIVE(vm->stringClass, "iteratorValue(_)", string_iteratorValue);
diff --git a/src/vm/wren_value.c b/src/vm/wren_value.c
index 06bfc85b..e715fd8d 100644
--- a/src/vm/wren_value.c
+++ b/src/vm/wren_value.c
@@ -854,16 +854,16 @@ Value wrenStringCodePointAt(WrenVM* vm, ObjString* string, uint32_t index)
 }
 
 // Uses the Boyer-Moore-Horspool string matching algorithm.
-uint32_t wrenStringFind(ObjString* haystack, ObjString* needle, uint32_t startIndex)
+uint32_t wrenStringFind(ObjString* haystack, ObjString* needle, uint32_t start)
 {
-  // Corner case, an empty needle is always found.
-  if (needle->length == 0) return 0;
+  // Edge case: An empty needle is always found.
+  if (needle->length == 0) return start;
 
-  // If the needle is longer than the haystack it won't be found.
-  if (needle->length > (haystack->length - startIndex)) return UINT32_MAX;
+  // If the needle goes past the haystack it won't be found.
+  if (start + needle->length > haystack->length) return UINT32_MAX;
 
   // If the startIndex is too far it also won't be found.
-  if (startIndex >= haystack->length) return UINT32_MAX;
+  if (start >= haystack->length) return UINT32_MAX;
 
   // Pre-calculate the shift table. For each character (8-bit value), we
   // determine how far the search window can be advanced if that character is
@@ -893,18 +893,18 @@ uint32_t wrenStringFind(ObjString* haystack, ObjString* needle, uint32_t startIn
   // Slide the needle across the haystack, looking for the first match or
   // stopping if the needle goes off the end.
   char lastChar = needle->value[needleEnd];
-  uint32_t range = (haystack->length - startIndex) - needle->length;
+  uint32_t range = haystack->length - needle->length;
 
-  for (uint32_t index = 0; index <= range; )
+  for (uint32_t index = start; index <= range; )
   {
     // Compare the last character in the haystack's window to the last character
     // in the needle. If it matches, see if the whole needle matches.
-    char c = haystack->value[startIndex + (index + needleEnd)];
+    char c = haystack->value[index + needleEnd];
     if (lastChar == c &&
-        memcmp(haystack->value + startIndex + index, needle->value, needleEnd) == 0)
+        memcmp(haystack->value + index, needle->value, needleEnd) == 0)
     {
       // Found a match.
-      return index + startIndex;
+      return index;
     }
 
     // Otherwise, slide the needle forward.
diff --git a/test/core/string/index_of.wren b/test/core/string/index_of.wren
index c6b3e02d..22fb8f79 100644
--- a/test/core/string/index_of.wren
+++ b/test/core/string/index_of.wren
@@ -5,12 +5,6 @@ System.print("abcd".indexOf("abcd")) // expect: 0
 System.print("abcd".indexOf("abcde")) // expect: -1
 System.print("abab".indexOf("ab")) // expect: 0
 
-System.print("abcd".indexOf("cd", 0)) // expect: 2
-System.print("abcd".indexOf("cd", 1)) // expect: 2
-System.print("abcd".indexOf("cd", 2)) // expect: 2
-System.print("abcd".indexOf("cd", 3)) // expect: -1
-System.print("abcd".indexOf("cd", 10)) // expect: -1
-
 // More complex cases.
 System.print("abcdefabcdefg".indexOf("defg")) // expect: 9
 System.print("abcdabcdabcd".indexOf("dab")) // expect: 3
diff --git a/test/core/string/index_of_start.wren b/test/core/string/index_of_start.wren
new file mode 100644
index 00000000..c850e82e
--- /dev/null
+++ b/test/core/string/index_of_start.wren
@@ -0,0 +1,39 @@
+// An empty string is anywhere you look for it.
+System.print("abcd".indexOf("", 0)) // expect: 0
+System.print("abcd".indexOf("", 1)) // expect: 1
+System.print("abcd".indexOf("", 2)) // expect: 2
+
+// Overlapping results.
+System.print("aaaaa".indexOf("aaaa", 0)) // expect: 0
+System.print("aaaaa".indexOf("aaaa", 1)) // expect: 1
+System.print("aaaaa".indexOf("aaaa", 2)) // expect: -1
+
+// It's OK if the needle extends past the end.
+System.print("abcd".indexOf("abcde", 0)) // expect: -1
+System.print("abcd".indexOf("cde", 3)) // expect: -1
+
+System.print("abcd".indexOf("cd", 0)) // expect: 2
+System.print("abcd".indexOf("cd", 1)) // expect: 2
+System.print("abcd".indexOf("cd", 2)) // expect: 2
+System.print("abcd".indexOf("cd", 3)) // expect: -1
+
+// Negative start.
+System.print("abcd".indexOf("cd", -4)) // expect: 2
+System.print("abcd".indexOf("cd", -3)) // expect: 2
+System.print("abcd".indexOf("cd", -2)) // expect: 2
+System.print("abcd".indexOf("cd", -1)) // expect: -1
+
+// Skips past earlier results.
+System.print("here as well as here".indexOf("here", 1)) // expect: 16
+
+// Non-ASCII. Note that it returns byte indices, not code points.
+System.print("søméஃthîng".indexOf("e", 2)) // expect: -1
+System.print("søméஃthîng".indexOf("m", 2)) // expect: 3
+System.print("søméஃthîng".indexOf("thî", 8)) // expect: 9
+
+// 8-bit clean.
+System.print("a\0b\0c".indexOf("\0", 0)) // expect: 1
+System.print("a\0b\0c".indexOf("a", 0)) // expect: 0
+System.print("a\0b\0c".indexOf("b\0c", 1)) // expect: 2
+System.print("a\0b\0c".indexOf("a\0b\0c\0d", 0)) // expect: -1
+System.print("a\0b\0a\0b".indexOf("a\0b", 0)) // expect: 0
diff --git a/test/core/string/index_of_start_not_int.wren b/test/core/string/index_of_start_not_int.wren
new file mode 100644
index 00000000..45897412
--- /dev/null
+++ b/test/core/string/index_of_start_not_int.wren
@@ -0,0 +1 @@
+"abcd".indexOf("bc", 12.34) // expect runtime error: Start must be an integer.
diff --git a/test/core/string/index_of_start_not_num.wren b/test/core/string/index_of_start_not_num.wren
new file mode 100644
index 00000000..9b3d1a23
--- /dev/null
+++ b/test/core/string/index_of_start_not_num.wren
@@ -0,0 +1,2 @@
+"abcd".indexOf("bc", "not num") // expect runtime error: Start must be a number.
+System.print("after")
\ No newline at end of file
diff --git a/test/core/string/index_of_start_too_large.wren b/test/core/string/index_of_start_too_large.wren
new file mode 100644
index 00000000..3edb3ac1
--- /dev/null
+++ b/test/core/string/index_of_start_too_large.wren
@@ -0,0 +1 @@
+"abcd".indexOf("bc", 4) // expect runtime error: Start out of bounds.
diff --git a/test/core/string/index_of_start_too_small.wren b/test/core/string/index_of_start_too_small.wren
new file mode 100644
index 00000000..ae0cac2d
--- /dev/null
+++ b/test/core/string/index_of_start_too_small.wren
@@ -0,0 +1 @@
+"abcd".indexOf("bc", -5) // expect runtime error: Start out of bounds.