diff --git a/assets/parser.c.erb b/assets/parser.c.erb
index ded8ba7..dd9c179 100644
--- a/assets/parser.c.erb
+++ b/assets/parser.c.erb
@@ -3,6 +3,17 @@
 #include <stdlib.h>
 #include <string.h>
 
+/**************************************************************************
+ * Public data
+ *************************************************************************/
+
+/** Token names. */
+const char * <%= @grammar.prefix %>token_names[] = {
+<% @grammar.tokens.each_with_index do |token, index| %>
+    "<%= token.name %>",
+<% end %>
+};
+
 /**************************************************************************
  * User code blocks
  *************************************************************************/
diff --git a/assets/parser.h.erb b/assets/parser.h.erb
index 3b32836..1c2a95a 100644
--- a/assets/parser.h.erb
+++ b/assets/parser.h.erb
@@ -116,6 +116,13 @@ typedef struct
     size_t user_terminate_code;
 } <%= @grammar.prefix %>context_t;
 
+/**************************************************************************
+ * Public data
+ *************************************************************************/
+
+/** Token names. */
+extern const char * <%= @grammar.prefix %>token_names[];
+
 void <%= @grammar.prefix %>context_init(<%= @grammar.prefix %>context_t * context, uint8_t const * input, size_t input_length);
 
 size_t <%= @grammar.prefix %>decode_code_point(uint8_t const * input, size_t input_length,
diff --git a/doc/user_guide.md b/doc/user_guide.md
index cf539cb..53adf5d 100644
--- a/doc/user_guide.md
+++ b/doc/user_guide.md
@@ -726,6 +726,28 @@ if (p_parse(&context) == P_USER_TERMINATED)
 }
 ```
 
+##> Data
+
+### `p_token_names`
+
+The `p_token_names` array contains the grammar-specified token names.
+It is indexed by the token ID.
+
+C example:
+
+```
+p_context_t context;
+p_context_init(&context, input, input_length);
+size_t result = p_parse(&context);
+if (p_parse(&context) == P_UNEXPECTED_TOKEN)
+{
+    p_position_t error_position = p_position(&context);
+    fprintf(stderr, "Error: unexpected token `%s' at row %u column %u\n",
+        p_token_names[context.token],
+        error_position.row + 1, error_position.col + 1);
+}
+```
+
 #> License
 
 Propane is licensed under the terms of the MIT License:
diff --git a/spec/test_error_positions.c b/spec/test_error_positions.c
index b871449..15aaf01 100644
--- a/spec/test_error_positions.c
+++ b/spec/test_error_positions.c
@@ -35,5 +35,8 @@ int main()
     assert(p_position(&context).row == 5);
     assert(p_position(&context).col == 4);
 
+    assert(strcmp(p_token_names[TOKEN_a], "a") == 0);
+    assert(strcmp(p_token_names[TOKEN_num], "num") == 0);
+
     return 0;
 }
diff --git a/spec/test_error_positions.d b/spec/test_error_positions.d
index 0282252..352b799 100644
--- a/spec/test_error_positions.d
+++ b/spec/test_error_positions.d
@@ -34,4 +34,7 @@ unittest
     p_context_init(&context, input);
     assert(p_parse(&context) == P_DECODE_ERROR);
     assert(p_position(&context) == p_position_t(5, 4));
+
+    assert(p_token_names[TOKEN_a] == "a");
+    assert(p_token_names[TOKEN_num] == "num");
 }