diff --git a/README.asciidoc b/README.asciidoc
index d8cb6e0961..a0d5bdcdbd 100644
--- a/README.asciidoc
+++ b/README.asciidoc
@@ -53,10 +53,10 @@ Variants:
  - `proc findAll(...)` returns a `seq[string]`
 
 [[proc-split]]
-==== split(string, Regex, maxsplit = -1): seq[string]
+==== split(string, Regex, maxsplit = -1, start = 0): seq[string]
 
 Splits the string with the given regex. This works according to the rules that
-Perl and Javascript use.
+Perl and Javascript use:
 
   - If the match is zero-width, then the string is still split:
     `"123".split(r"") == @["1", "2", "3"]`.
@@ -66,6 +66,8 @@ Perl and Javascript use.
     times. This means that there will be `maxsplit` strings in the output seq.
     `"1.2.3".split(re"\.", maxsplit = 2) == @["1", "2.3"]`
 
+`start` behaves the same as in link:#proc-find[`find(...)`].
+
 [[proc-replace]]
 ==== replace(string, Regex, sub): string
 
diff --git a/src/nre.nim b/src/nre.nim
index 47e374978b..6743a82dcb 100644
--- a/src/nre.nim
+++ b/src/nre.nim
@@ -384,13 +384,13 @@ proc renderBounds(str: string, bounds: Slice[int]): string =
   for i in bounds.a .. bounds.b:
     result.add("^")
 
-proc split*(str: string, pattern: Regex, maxSplit = -1): seq[string] =
+proc split*(str: string, pattern: Regex, maxSplit = -1, start = 0): seq[string] =
   result = @[]
-  var lastIdx = 0
+  var lastIdx = start
   var splits = 0
   var bounds: Slice[int]
 
-  for match in str.findIter(pattern):
+  for match in str.findIter(pattern, start = start):
     # upper bound is exclusive, lower is inclusive:
     #
     # 0123456
@@ -401,7 +401,7 @@ proc split*(str: string, pattern: Regex, maxSplit = -1): seq[string] =
     # "12".split("") would be @["", "1", "2"], but
     # if we skip an empty first match, it's the correct
     # @["1", "2"]
-    if bounds.a < bounds.b or bounds.a > 0:
+    if bounds.a < bounds.b or bounds.a > start:
       result.add(str.substr(lastIdx, bounds.a - 1))
       splits += 1
 
diff --git a/test/split.nim b/test/split.nim
index 7aed714a4d..62b88427b2 100644
--- a/test/split.nim
+++ b/test/split.nim
@@ -43,3 +43,8 @@ suite "string splitting":
     check("a.b"                          .split(re"\.")    .join(",") == "a,b")
     check(""                             .split(re"")      .len       == 0)
     check(":"                            .split(re"")      .len       == 1)
+
+  test "start position":
+    check("abc".split(re"", start = 1) == @["b", "c"])
+    check("abc".split(re"", start = 2) == @["c"])
+    check("abc".split(re"", start = 3) == newSeq[string]())