diff --git a/NEWS.md b/NEWS.md
index 7cf63f0b7..d781f90e1 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -40,6 +40,8 @@
 
 5. Non-equi joins combining an equality condition with two inequality conditions on the same column (e.g., `on = .(id == id, val >= lo, val <= hi)`) no longer error, [#7641](https://github.com/Rdatatable/data.table/issues/7641). The internal `chmatchdup` remapping of duplicate `rightcols` was overwriting the original column indices, causing downstream code to reference non-existent columns. Thanks @tarun-t for the report and fix, and @aitap for the diagnosis.
 
+6. `frollapply()` could have produce output longer than the input when the window-length is also longer than the input [#7646](https://github.com/Rdatatable/data.table/issues/7646). Thanks to @hadley-johnson for reporting and @jangorecki for the fix.
+
 ### Notes
 
 1. {data.table} now depends on R 3.5.0 (2018).
diff --git a/R/frollapply.R b/R/frollapply.R
index 9c1b0c26d..6a138547d 100644
--- a/R/frollapply.R
+++ b/R/frollapply.R
@@ -278,8 +278,11 @@ frollapply = function(X, N, FUN, ..., by.column=TRUE, fill=NA, align=c("right","
     cpy = copy
     ansMask = function(len, n) {
       mask = rep(TRUE, len)
-      if (n) ## handle n==0
+      if (n) { ## handle n==0
+        if (n > len)
+          n = len + 1L ## bugfix #7646
         mask[seq_len(n-1L)] = FALSE
+      }
       mask
     }
     tight0 = function(i, dest, src, n) FUN(dest, ...) ## skip memcpy when n==0
diff --git a/inst/tests/froll.Rraw b/inst/tests/froll.Rraw
index 5f8225910..22fd88423 100644
--- a/inst/tests/froll.Rraw
+++ b/inst/tests/froll.Rraw
@@ -2317,3 +2317,6 @@ test(6015.907, frolladapt(c(1L,3L,2L), 2L), error="be sorted, have no duplicates
 test(6015.908, frolladapt(c(1L,2L,2L), 2L), error="be sorted, have no duplicates, have no NAs")
 test(6015.909, frolladapt(c(1L,2L,NA_integer_), 2L), error="be sorted, have no duplicates, have no NAs") ## loop that checks for sorted will detect NAs as well, except for first element
 test(6015.910, frolladapt(c(NA_integer_,1L,2L), 2L), error="be sorted, have no duplicates, have no NAs") ## first NA is detected by extra check
+
+# frollapply can produce output longer than the input when the window-length is also longer than the input #7646
+test(6100.1, frollapply(c(1,2,3,4,5),N=7,FUN=sum), rep(NA, 5L))