diff --git a/tests/test_tokenize.py b/tests/test_tokenize.py
index 594fe99..332c0e3 100644
--- a/tests/test_tokenize.py
+++ b/tests/test_tokenize.py
@@ -177,7 +177,8 @@
     assert (token1 == token2) == equals
 
 
-# Draws a token with the values split_tokens outputs
+# Draws a token using an existing strategy but with a blank location
+# and unknown type, just like split_tokens outputs
 @composite
 def draw_token_splitted(draw, strategy):
     token = draw(strategy)
@@ -187,6 +188,7 @@
 
 
 # Generates an alternating sequence of unknown or whitespace tokens
+# intended for splitting in to separate tokens
 @composite
 def draw_tokens_to_split(draw):
     source = ""
@@ -195,6 +197,7 @@
     drawing_whitespace = draw(booleans())
     for _ in elements:
         if drawing_whitespace:
+            # Multiple whitespaces get split in to multiple tokens
             strategy = one_of([draw_token_space(), draw_token_newline()])
             locationed = draw_token_splitted(strategy)
             tokens += draw(lists(locationed, min_size=1))
@@ -209,6 +212,14 @@
 
 
 # Test that the tokenizer can split tokens properly
+# We expect the following behaviour:
+# - Whitespace is a tab, space or new line
+# - Non-whitespace is anything else
+# - Whitespace and non-whitespace are separated in to tokens
+# - Whitespace characters are split in to multiple tokens
+# - Non-whitespace characters are combined in to a single token
+# - Each token type is UNKNOWN
+# - Each token location is line 1 column 1 of file ""
 @given(draw_tokens_to_split())
 def test_tokenize_split_tokens(test_data):
     (source, tokens) = test_data
@@ -236,6 +247,15 @@
 
 
 # Test that the tokenizer can determine locations
+# We expect the following behaviour:
+# - Only the token location is modified
+# - Each token's location filename is the generated filename
+# - A token's line is equal to 1 plus the number of tokens with the
+#   type NEWLINE before the token
+# - A token's column is equal to 1 plus the sum of previous token's
+#   value lengths 'before' it, where 'before' is defined as any
+#   token between the last NEWLINE (or start of file) before the token
+#   and the token itself
 @given(draw_tokens_locations())
 def test_tokenize_locations(test_data):
     (input, located, filename) = test_data
@@ -243,6 +263,7 @@
 
 
 # Draws a token and possibly adds garbage
+# This is to ensure that tokens must completely match a value
 @composite
 def draw_token_classified_garbled(draw):
     token = draw(draw_token_classified())
@@ -271,13 +292,22 @@
 
 
 # Test that classification can add types properly
+# We expect the following behaviour:
+# - Only the token type is modified
+# - The token's type is set based on which following condition its value meets:
+#   - KEYWORD if it entirely matches a keyword
+#   - BOOL if it is True or False
+#   - WHITESPACE if it is a space or tab character
+#   - NEWLINE if it is a new line character
+#   - SHEBANG if it starts with a #1
+#   - UNKNOWN if it is none of the above
 @given(draw_tokens_classified())
 def test_tokenize_classification(test_data):
     (input, tokens) = test_data
     assert tokenize.classify_tokens(input) == tokens
 
 
-# Draw random source code that might be invalid for fuzzing
+# Draw a random string made of token values
 @composite
 def draw_source_fuzz(draw):
     tokens = draw(lists(draw_token_classified()))
@@ -288,6 +318,10 @@
 
 
 # Test that the tokenize function behaves as we expect
+# We expect the following behaviour:
+# - The tokenizer splits the tokens as expected, then
+# - The tokenizer classifies tokens as expected, then
+# - The tokenizer sets the token locations as expected
 @given(draw_source_fuzz(), text())
 def test_tokenize_fuzz(source, filename):
     split = tokenize.split_tokens(source)