diff --git a/README.md b/README.md
index 28a4f2e..840b0ef 100644
--- a/README.md
+++ b/README.md
@@ -185,6 +185,7 @@ Shell completion scripts for Bash, Fish and Zsh can be found in respective subdi
 - It's  advisable  to copy URLs directly from the browser address bar, i.e., along with the leading `http://` or `https://` token. buku looks up title data (found within <title></title> tags of HTML) from the web ONLY for fully-formed HTTP(S) URLs.
 - If the URL contains characters like `;`, `&` or brackets they may be interpreted specially by the shell. To avoid it, add the URL within single or double (`'`/`"`) quotes.
 - URLs are unique in DB. The same URL cannot be added twice. You can update tags and re-fetch title data.
+- For tags, comma (`,`) is the delimiter in DB. Tags are comma separated, filtered (for unique tags) and sorted. Hence, a tag cannot have comma(s) in it. In-tag commas are replaced by spaces.
 - **Update** operation:
   - If --title, --tag or --comment is passed without argument, clear the corresponding field from DB.
   - If --url is passed (and --title is omitted), update the title from web using the URL.
diff --git a/buku b/buku
index 0a555ca..b925596 100755
--- a/buku
+++ b/buku
@@ -893,9 +893,9 @@ def parse_tags(keywords=[]):
                 tags += ','
 
         if tag[-1] == ',':          # if delimiter is present, maintain it (e.g. token1, token2)
-            tag = tag.strip(',') + ','
+            tag = tag.strip(',').replace(',', ' ') + ','
         else:                       # a token in a multi-word tag (e.g. token1 token2)
-            tag = tag.strip(',')
+            tag = tag.strip(',').replace(',', ' ')
 
         if tag == ',':              # isolated delimiter (e.g. token1 , token2)
             if tags[-1] != ',':
diff --git a/buku.1 b/buku.1
index 28067fb..4f7cf16 100644
--- a/buku.1
+++ b/buku.1
@@ -30,6 +30,8 @@ If the URL contains characters like ';', '&' or brackets they may be interpreted
 .PP
 URLs are unique in DB. The same URL cannot be added twice. You can update tags and re-fetch title data.
 .PP
+For tags, comma (',') is the delimiter in DB. Tags are comma separated, filtered (for unique tags) and sorted. Hence, a tag cannot have comma(s) in it. In-tag commas are replaced by spaces.
+.PP
 \fBUpdate\fR operation:
   - If --title, --tag or --comment is passed without argument, clear the corresponding field from DB.
   - If --url is passed (and --title is omitted), update the title from web using the URL.