diff --git a/src/ast/link.rs b/src/ast/link.rs
index eddb2ef..48f55d2 100644
--- a/src/ast/link.rs
+++ b/src/ast/link.rs
@@ -1,7 +1,7 @@
 use rowan::ast::{support, AstNode};
 
 use super::{AffiliatedKeyword, Link, Paragraph, Token};
-use crate::syntax::SyntaxKind;
+use crate::{syntax::SyntaxKind, SyntaxElement};
 
 impl Link {
     /// Returns link destination
@@ -35,9 +35,59 @@ impl Link {
     /// assert!(!link.has_description());
     /// let link = Org::parse("[[https://google.com][Google]]").first_node::<Link>().unwrap();
     /// assert!(link.has_description());
+    /// let link = Org::parse("[[https://example.com][*abc* /abc/]]").first_node::<Link>().unwrap();
+    /// assert!(link.has_description());
     /// ```
     pub fn has_description(&self) -> bool {
-        support::token(self.syntax(), SyntaxKind::TEXT).is_some()
+        self.syntax()
+            .children_with_tokens()
+            .any(|e| e.kind() == SyntaxKind::L_BRACKET)
+    }
+
+    /// Returns parsed description
+    ///
+    /// Returns empty iterator if this link doesn't contain description
+    ///
+    /// ```rust
+    /// use orgize::{Org, ast::Link, SyntaxKind};
+    ///
+    /// let link = Org::parse("[[https://google.com]]").first_node::<Link>().unwrap();
+    /// assert_eq!(link.description().count(), 0);
+    ///
+    /// let link = Org::parse("[[https://google.com][Google]]").first_node::<Link>().unwrap();
+    /// let description = link.description().collect::<Vec<_>>();
+    /// assert_eq!((description[0].kind(), description[0].to_string()), (SyntaxKind::TEXT, "Google".into()));
+    ///
+    /// let link = Org::parse("[[https://example.com][*abc* /abc/]]").first_node::<Link>().unwrap();
+    /// let description = link.description().collect::<Vec<_>>();
+    /// assert_eq!((description[0].kind(), description[0].to_string()), (SyntaxKind::BOLD, "*abc*".into()));
+    /// assert_eq!((description[2].kind(), description[2].to_string()), (SyntaxKind::ITALIC, "/abc/".into()));
+    /// ```
+    pub fn description(&self) -> impl Iterator<Item = SyntaxElement> {
+        self.syntax()
+            .children_with_tokens()
+            .skip_while(|e| e.kind() != SyntaxKind::L_BRACKET)
+            .skip(1)
+            .take_while(|e| e.kind() != SyntaxKind::R_BRACKET2)
+    }
+
+    /// Returns description raw string
+    ///
+    /// Returns empty string if this link doesn't contain description
+    ///
+    /// ```rust
+    /// use orgize::{Org, ast::Link};
+    ///
+    /// let link = Org::parse("[[https://google.com]]").first_node::<Link>().unwrap();
+    /// assert_eq!(link.description_raw(), "");
+    /// let link = Org::parse("[[https://google.com][Google]]").first_node::<Link>().unwrap();
+    /// assert_eq!(link.description_raw(), "Google");
+    /// let link = Org::parse("[[https://example.com][*abc* /abc/]]").first_node::<Link>().unwrap();
+    /// assert_eq!(link.description_raw(), "*abc* /abc/");
+    /// ```
+    pub fn description_raw(&self) -> String {
+        self.description()
+            .fold(String::new(), |acc, e| acc + &e.to_string())
     }
 
     /// Returns `true` if link is an image link