diff --git a/README.md b/README.md
index a94efdf..2b4497e 100644
--- a/README.md
+++ b/README.md
@@ -359,33 +359,37 @@ It is compatible with heterogeneous types. I.e.,
 #### `as_list`
 
 By default, when `read` is [called with multiple types](#multiple-tokens),
-it returns a `list` with the requested tokens. If, for some reason, you
-don't need implicit conversion to a `list`, `read` can return a
-generator instead. This is achieved by passing `as_list=False`, as in
-`read(float, amount=10000, as_list=False)`.
+it returns a generator with the requested tokens. If you want the result
+as a `list` instead, pass `as_list=True`, as in
+`read(float, amount=10000, as_list=True)`.
 
-This has the advantage of avoiding unnecessary iteration to create the
-list and copy the tokens into it. For example, you could use it to create
-a large `numpy` array: 
+When reading a single token (one type with `amount=1`), `read` normally
+returns the token directly (e.g., an `int` or `str`). Passing
+`as_list=True` will wrap the result in a list, e.g.,
+`read(int, as_list=True)` returns `[5]` instead of `5`.
+
+Using `as_list=False` (the default) has the advantage of avoiding
+unnecessary iteration to create the list and copy the tokens into it.
+For example, you could use it to create a large `numpy` array:
 
 ```python
 import numpy as np
 from easyinput import read
 
-arr = np.array(read(float, amount=1000000, as_list=False))
+arr = np.array(read(float, amount=1000000))
 ```
 
-In the previous block of code, only one iteration through all of 
+In the previous block of code, only one iteration through all of
 the tokens takes place—the one to fill the `np.array`—
 whilst if `as_list` were set to
-`False`, two iterations would take place: one to fill the list, and
+`True`, two iterations would take place: one to fill the list, and
 another to fill the array.
 
 Keep in mind that the generator returned by `read` when `as_list` is
-set to `True` *does not consume any input until you iterate through it*.
+set to `False` *does not consume any input until you iterate through it*.
 This is the reason for which `read_many` does **not** support this
 keyword argument: it *must* consume input in order to discern
-whether it should keep going or not, so it always returns 
+whether it should keep going or not, so it always returns
 multiple-token queries as lists.
 
 <br>
@@ -683,7 +687,7 @@ a.say_word()
 
 ## Interface
 
-- *(function)* **`easyinput.read(*types, amount=1, file=_StdIn, as_list=True)`** <br>
+- *(function)* **`easyinput.read(*types, amount=1, file=_StdIn, as_list=False)`** <br>
 Returns tokens from the input stream specified by `file`.
 `_StdIn` is an alias for standard input. 
     
@@ -705,10 +709,13 @@ Returns tokens from the input stream specified by `file`.
     (e.g. `io.StringIO`) that specifies where to read input from.
     
     - `as_list: bool` <br>
-    If `True`, if more than one token is requested, converted
-    tokens are returned as a list, and thus input is
-    immediately consumed. Otherwise, a generator is
-    returned (and no input is consumed).
+    If `True`, the result is returned as a list. For single-token
+    reads, this wraps the token in a list (e.g., `[5]` instead of `5`).
+    For multiple-token reads, converted tokens are returned as a list,
+    and thus input is immediately consumed.
+    If `False` (the default), single-token reads return the token
+    directly, and multiple-token reads return a generator
+    (no input is consumed until iterated).
 
 - *(function)* **`easyinput.read_many(*types, amount=1, file=_StdIn)`** <br>
 Returns a generator that yields tokens from the input stream
diff --git a/easyinput/tokenizer.py b/easyinput/tokenizer.py
index 00e0744..66ed0fe 100644
--- a/easyinput/tokenizer.py
+++ b/easyinput/tokenizer.py
@@ -112,18 +112,21 @@ def multiple_tokens(self, amount, type1=str, *types):
 _StdIn = StdIn()
 
 
-@kwd_only(file=_StdIn, amount=1, as_list=True)  # Python 2 compatibility
+@kwd_only(file=_StdIn, amount=1, as_list=False)  # Python 2 compatibility
 def read(*types, **kwargs):
     """
     Py3 signature:
-        `read(*types, file:iter=_StdIn, amount:int=1, as_list:bool=True)`
+        `read(*types, file:iter=_StdIn, amount:int=1, as_list:bool=False)`
     This function returns one or more tokens converted to
     the types specified by *types.
     This is the main function in the module.
     """
     tokens, amount, as_list = _get_tokenizer(kwargs), _get_amount(kwargs), kwargs['as_list']
     method, args = _select_method(tokens, types, amount, as_list)
-    return method(*args)
+    result = method(*args)
+    if as_list and not isinstance(result, list):
+        return [result]
+    return result
 
 
 @kwd_only(file=_StdIn, amount=1)  # Python 2 compatibility
diff --git a/test/test_amount.py b/test/test_amount.py
index 3b15cab..0f61f19 100644
--- a/test/test_amount.py
+++ b/test/test_amount.py
@@ -33,3 +33,17 @@ def test_4():
     expected = ('a', 2, 'c', 3)
     tokens = easyinput.read(chr, int, amount=2, file=source)
     assert all(inp == val for inp, val in zip(tokens, expected))
+
+def test_5():
+    source = io.StringIO("5 4 ")
+    expected = [5]
+    tokens = easyinput.read(int, amount = 1, as_list = True, file=source)
+    assert isinstance(tokens, list)
+    assert tokens == expected
+    
+def test_6():
+    source = io.StringIO("hello good bye")
+    expected = ["hello"]
+    tokens = easyinput.read(str, amount = 1, as_list = True, file = source)
+    assert isinstance(tokens, list)
+    assert tokens == expected