diff --git a/stdlib/camlinternalFormat.ml b/stdlib/camlinternalFormat.ml
index 7c36e22afc71..d0114916e8c6 100644
--- a/stdlib/camlinternalFormat.ml
+++ b/stdlib/camlinternalFormat.ml
@@ -221,6 +221,9 @@ let default_float_precision = -6
      necessary".  For the other FP formats, we take the absolute value
      of the precision, hence 6 digits by default. *)
 
+(* Default precision for OCaml float printing (%F). *)
+let default_ocaml_float_precision = 12
+
 (******************************************************************************)
                                (* Externals *)
 
@@ -1408,9 +1411,8 @@ let format_of_iconvn = function
 
 (* Generate the format_float first argument form a float_conv. *)
 let format_of_fconv fconv prec =
-  if fconv = Float_F then "%.12g" else
     let prec = abs prec in
-    let symb = char_of_fconv fconv in
+    let symb = if fconv = Float_F then 'g' else char_of_fconv fconv in
     let buf = buffer_create 16 in
     buffer_add_char buf '%';
     bprint_fconv_flag buf fconv;
@@ -1730,10 +1732,13 @@ and make_float_padding_precision : type x y a b c d e f .
     ((b, c) acc -> f) -> (b, c) acc ->
     (a, b, c, d, e, f) fmt ->
     (x, y) padding -> (y, float -> a) precision -> float_conv -> x =
+  let default_float_precision fconv =
+    if fconv = Float_F then default_ocaml_float_precision
+    else default_float_precision in
   fun k acc fmt pad prec fconv -> match pad, prec with
   | No_padding, No_precision ->
     fun x ->
-      let str = convert_float fconv default_float_precision x in
+      let str = convert_float fconv (default_float_precision fconv) x in
       make_printf k (Acc_data_string (acc, str)) fmt
   | No_padding, Lit_precision p ->
     fun x ->
@@ -1745,7 +1750,7 @@ and make_float_padding_precision : type x y a b c d e f .
       make_printf k (Acc_data_string (acc, str)) fmt
   | Lit_padding (padty, w), No_precision ->
     fun x ->
-      let str = convert_float fconv default_float_precision x in
+      let str = convert_float fconv (default_float_precision fconv) x in
       let str' = fix_padding padty w str in
       make_printf k (Acc_data_string (acc, str')) fmt
   | Lit_padding (padty, w), Lit_precision p ->
@@ -1758,7 +1763,7 @@ and make_float_padding_precision : type x y a b c d e f .
       make_printf k (Acc_data_string (acc, str)) fmt
   | Arg_padding padty, No_precision ->
     fun w x ->
-      let str = convert_float fconv default_float_precision x in
+      let str = convert_float fconv (default_float_precision fconv) x in
       let str' = fix_padding padty w str in
       make_printf k (Acc_data_string (acc, str')) fmt
   | Arg_padding padty, Lit_precision p ->
diff --git a/stdlib/printf.mli b/stdlib/printf.mli
index a80aa04544ba..b1582f2d2c4f 100644
--- a/stdlib/printf.mli
+++ b/stdlib/printf.mli
@@ -110,8 +110,8 @@ val fprintf : out_channel -> ('a, out_channel, unit) format -> 'a
 
    The optional [precision] is a dot [.] followed by an integer
    indicating how many digits follow the decimal point in the [%f],
-   [%e], and [%E] conversions. For instance, [%.4f] prints a [float] with
-   4 fractional digits.
+   [%F], [%e], [%E], [%g], [%G], [%h], and [%H] conversions.
+   For instance, [%.4f] prints a [float] with 4 fractional digits.
 
    The integer in a [width] or [precision] can also be specified as
    [*], in which case an extra integer argument is taken to specify