diff --git a/R/fmelt.R b/R/fmelt.R
index f8cb42b3e..3e6f18444 100644
--- a/R/fmelt.R
+++ b/R/fmelt.R
@@ -22,6 +22,9 @@ melt.default = function(data, ..., na.rm = FALSE, value.name = "value") {
 patterns = function(..., cols=character(0L), ignore.case=FALSE, perl=FALSE, fixed=FALSE, useBytes=FALSE) {
   # if ... has no names, names(list(...)) will be "";
   #   this assures they'll be NULL instead
+  if (!is.character(cols) || anyNA(cols)) {
+    stopf("cols must be a character vector of column names")
+  }
   L = list(...)
   p = unlist(L, use.names = any(nzchar(names(L))))
   if (!is.character(p))
diff --git a/inst/tests/tests.Rraw b/inst/tests/tests.Rraw
index ff24b1f8d..216674504 100644
--- a/inst/tests/tests.Rraw
+++ b/inst/tests/tests.Rraw
@@ -12428,6 +12428,11 @@ DTout = data.table(
 )
 test(1866.6, melt(DT, measure.vars = patterns("^x", "^y", cols=names(DT))), DTout)
 
+# informative errors for bad user-provided cols arg to patterns
+DT = data.table(x1=1,x2=2,y1=3,y2=4)
+test(1866.80, melt(DT, measure.vars=patterns("2", cols=NULL)), error="cols must be a character vector of column names")
+test(1866.81, melt(DT, measure.vars=patterns("2", cols=NA_character_)), error="cols must be a character vector of column names")
+
 # auto fill too few column names (#1625) and auto fill=TRUE when too many column names
 test(1867.01, fread("A,B\n1,3,5,7\n2,4,6,8\n"), data.table(A=1:2, B=3:4, V3=5:6, V4=7:8),
               warning="Detected 2 column names but.*4.*Added 2 extra default column names at the end[.]")