diff --git a/DESCRIPTION b/DESCRIPTION
index ceaa19d1..4222f225 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -26,4 +26,4 @@ Collate: 'dimensions.r' 'helper-arrange.r' 'helper-col-wise.r'
     'progress.r' 'quote.r' 'rbind-matrix.r' 'rbind.r'
     'simplify-array.r' 'simplify-data-frame.r' 'simplify-vector.r'
     'split-array.r' 'split-data-frame.r' 'split-indices.r' 'split.r'
-    'utils.r'
+    'utils.r' 'loop-apply.r'
diff --git a/NEWS b/NEWS
index 3c1424b5..1c9cf8a2 100644
--- a/NEWS
+++ b/NEWS
@@ -1,6 +1,7 @@
 plyr 1.2 (2010-XX-XX) ---------------------------------------------------
 
 * llply gains .parallel argument that when set to true will apply parallel functions in parallel using foreach package
+* llply: in serial case, for loop replaced with custom C function that takes about 40% less time (or about 20% less time than lapply)
 
 plyr 1.1 (2010-07-19) ---------------------------------------------------
 
diff --git a/R/loop-apply.r b/R/loop-apply.r
new file mode 100644
index 00000000..f42b6103
--- /dev/null
+++ b/R/loop-apply.r
@@ -0,0 +1,12 @@
+#' Split indices.
+#' An optimised version of split for the special case of splitting row
+#' indices into groups, as used by \code{\link{splitter_d}}
+#'
+#' @param index integer indices
+#' @param group integer groups
+#' @param n largest integer (may not appear in index)
+#' @useDynLib plyr
+#' @keywords internal manip
+loop_apply <- function(n, f, env = parent.frame()) {
+  .Call("loop_apply", as.integer(n), f, env)
+}
\ No newline at end of file
diff --git a/R/ply-list.r b/R/ply-list.r
index d2326f9d..df20fb0a 100644
--- a/R/ply-list.r
+++ b/R/ply-list.r
@@ -75,7 +75,7 @@ llply <- function(.data, .fun = NULL, ..., .progress = "none", .inform = FALSE,
     }
     result <- foreach(i = seq_len(n)) %dopar% do.ply(i)
   } else {
-    result <- lapply(seq_len(n), do.ply)
+    result <- loop_apply(n, do.ply)
   }
   
   attributes(result)[c("split_type", "split_labels")] <-
diff --git a/R/split-indices.r b/R/split-indices.r
index 368afd2c..12d4557e 100644
--- a/R/split-indices.r
+++ b/R/split-indices.r
@@ -10,5 +10,4 @@
 split_indices <- function(index, group, n = max(group)) {
   if (length(index) == 0) return(integer())
   .Call("split_indices", index, group, as.integer(n))
-}
-  
\ No newline at end of file
+}
\ No newline at end of file
diff --git a/benchmark/timings.csv b/benchmark/timings.csv
index 04701d22..f03147e1 100644
--- a/benchmark/timings.csv
+++ b/benchmark/timings.csv
@@ -45,3 +45,17 @@
 "bench-dlply.r","6i-a",2010-07-20 13:46:52,"67da54f",0.248,0.033,0.281
 "bench-dlply.r","6i-b",2010-07-20 13:46:52,"67da54f",0.631,0.034,0.666
 "bench-dlply.r","6i-c",2010-07-20 13:46:57,"67da54f",4.123,0.072,4.229
+"bench-llply.r","1e5",2010-07-20 15:17:10,"6562dda",0.461,0.035,0.497
+"bench-llply.r","1e6",2010-07-20 15:17:16,"6562dda",5.668,0.047,5.718
+"bench-llply.r","1e5",2010-07-20 15:17:18,"6562dda",0.461,0.042,0.503
+"bench-llply.r","1e6",2010-07-20 15:17:23,"6562dda",5.591,0.025,5.621
+"bench-dlply.r","6-a",2010-07-20 15:17:29,"6562dda",1.505,0.415,1.921
+"bench-dlply.r","6-b",2010-07-20 15:17:35,"6562dda",3.503,2.072,5.58
+"bench-dlply.r","6i-a",2010-07-20 15:17:35,"6562dda",0.251,0.032,0.283
+"bench-dlply.r","6i-b",2010-07-20 15:17:36,"6562dda",0.65,0.048,0.697
+"bench-dlply.r","6i-c",2010-07-20 15:17:40,"6562dda",4.182,0.07,4.255
+"bench-dlply.r","6-a",2010-07-20 15:17:45,"6562dda",1.169,0.375,1.545
+"bench-dlply.r","6-b",2010-07-20 15:17:51,"6562dda",3.524,1.959,5.484
+"bench-dlply.r","6i-a",2010-07-20 15:17:51,"6562dda",0.247,0.029,0.276
+"bench-dlply.r","6i-b",2010-07-20 15:17:52,"6562dda",0.638,0.031,0.67
+"bench-dlply.r","6i-c",2010-07-20 15:17:56,"6562dda",4.132,0.062,4.196
diff --git a/src/loop-apply.c b/src/loop-apply.c
new file mode 100644
index 00000000..59588b0e
--- /dev/null
+++ b/src/loop-apply.c
@@ -0,0 +1,25 @@
+#include <R.h>
+#include <Rdefines.h>
+
+SEXP loop_apply(SEXP n, SEXP f, SEXP rho) {
+  if(!isFunction(f)) error("'f' must be a function");
+  if(!isEnvironment(rho)) error("'rho' should be an environment");
+
+  int n1 = INTEGER(n)[0];
+  
+  SEXP results, R_fcall;
+  PROTECT(results = allocVector(VECSXP, n1));
+  PROTECT(R_fcall = lang2(f, R_NilValue));
+
+  SEXP ii;
+  for(int i = 0; i < n1; i++) {
+    ii = PROTECT(ScalarInteger(i + 1));
+    SETCADR(R_fcall, ii);
+    SET_VECTOR_ELT(results, i, eval(R_fcall, rho));
+    
+    UNPROTECT(1);
+  }
+
+  UNPROTECT(2);
+  return results;
+}