diff --git a/hledger-lib/Hledger/Read/Common.hs b/hledger-lib/Hledger/Read/Common.hs
index ea3a1bc04..85dac541f 100644
--- a/hledger-lib/Hledger/Read/Common.hs
+++ b/hledger-lib/Hledger/Read/Common.hs
@@ -24,6 +24,7 @@ import Prelude.Compat hiding (readFile)
 import Control.Monad.Compat
 import Control.Monad.Except (ExceptT(..), runExceptT, throwError) --, catchError)
 import Control.Monad.State.Strict
+import Data.Char
 import Data.Data
 import Data.Default
 import Data.Functor.Identity
@@ -599,7 +600,7 @@ rawnumberp = do
     (firstSep, groups) <- option (Nothing, []) $ do
         leadingDigits <- some digitChar
         option (Nothing, [leadingDigits]) . try $ do
-            firstSep <- oneOf sepChars
+            firstSep <- oneOf sepChars <|> whitespaceChar
             groups <- some digitChar `sepBy1` char firstSep
             return (Just firstSep, leadingDigits : groups)
 
@@ -614,11 +615,14 @@ rawnumberp = do
         return (lastSep, fromMaybe [] digits)
 
     -- make sure we didn't leading part of mistyped number
-    notFollowedBy $ oneOf sepChars
+    notFollowedBy $ oneOf sepChars <|> (whitespaceChar >> digitChar)
 
     return $ dbg8 "rawnumberp" (firstSep, groups, extraGroup)
     <?> "rawnumberp"
 
+-- | Parse a unicode char that represents any non-control space char (Zs general category).
+whitespaceChar :: TextParser m Char
+whitespaceChar = charCategory Space
 
 -- test_numberp = do
 --       let s `is` n = assertParseEqual (parseWithState mempty numberp s) n
diff --git a/hledger-lib/hledger_journal.m4.md b/hledger-lib/hledger_journal.m4.md
index fb55737d7..9d0bb45e2 100644
--- a/hledger-lib/hledger_journal.m4.md
+++ b/hledger-lib/hledger_journal.m4.md
@@ -267,7 +267,8 @@ Amounts consist of a number and (usually) a currency symbol or commodity name. S
   `3 "green apples"`\
   `-$1,000,000.00`\
   `INR 9,99,99,999.00`\
-  `EUR -2.000.000,00`
+  `EUR -2.000.000,00`\
+  `1 999 999.9455`
 
 As you can see, the amount format is somewhat flexible:
 
@@ -275,7 +276,8 @@ As you can see, the amount format is somewhat flexible:
 - the commodity is a symbol, word, or phrase, on the left or right, with or without a separating space.
   If the commodity contains numbers, spaces or non-word punctuation it must be enclosed in double quotes.
 - negative amounts with a commodity on the left can have the minus sign before or after it
-- digit groups (thousands, or any other grouping) can be separated by commas (in which case period is used for decimal point) or periods (in which case comma is used for decimal point)
+- digit groups (thousands, or any other grouping) can be separated by space or comma or period and should be used as separator between all groups
+- decimal part can be separated by comma or period and should be different from digit groups separator
 
 You can use any of these variations when recording data. However, there is some ambiguous way of representing numbers like `$1.000` and `$1,000` both may mean either one thousand or one dollar. By default hledger will assume that this is sole delimiter is used only for decimals. On the other hand commodity format declared prior to that line will help to resolve that ambiguity differently:
 
diff --git a/tests/journal/numbers.test b/tests/journal/numbers.test
index b81669c95..01e169173 100644
--- a/tests/journal/numbers.test
+++ b/tests/journal/numbers.test
@@ -43,6 +43,32 @@ hledger bal -f -
 >>>
 >>>=1
 
+# Space between digits groups
+hledger bal -f - --no-total
+<<<
+2017/1/1
+	a   1 000.00 EUR
+	b  -1 000.00 EUR
+>>>
+        1 000.00 EUR  a
+       -1 000.00 EUR  b
+>>>2
+>>>=0
+
+# Space between digits groups in commodity directive
+hledger bal -f - --no-total
+<<<
+commodity 1 000.00 EUR
+
+2017/1/1
+	a   1,000.00 EUR
+	b  -1,000.00 EUR
+>>>
+        1 000.00 EUR  a
+       -1 000.00 EUR  b
+>>>2
+>>>=0
+
 # Default commodity
 hledger bal -f -
 <<<