range vs xrange in Python (What Changed in Python 3)

Range and xrange in Python come from two eras: in Python 2, range() built a full list while xrange() produced numbers lazily. Python 3 dropped xrange and made range work the lazy way, so today you just use range():

# Python 3
for i in range(1_000_000):   # lazy and memory-light, like the old xrange
    ...

Python 2 reached end of life in 2020, so I describe its behavior rather than run it. Every Python 3 example ran on Python 3.12.5 in Command Prompt.

I’ll show the NameError you get from old code, how much memory range saves, what a Python 3 range can do, how fast membership checks are, and how to port xrange code.

range vs xrange: what’s the difference?

	Python 2 range()	Python 2 xrange()	Python 3 range()
Returns	A list	An xrange object	A range object
Memory	Grows with size	Small and fixed	Small and fixed
Indexing, len()	Yes	Yes	Yes
Slicing	Yes (list)	No	Yes (returns a range)
Exists in Python 3	Renamed behavior	No (NameError)	Yes

In short, Python 3’s range is the old xrange with extra features, and the list-building version is gone. Wrap it in list() when you actually need a list.

Why does xrange give a NameError in Python 3?

Because it no longer exists. Old tutorials and Python 2 code still use it, and Python 3 stops at the first call:

try:
    for i in xrange(3):
        print(i)
except NameError as err:
    print("NameError:", err)

for i in range(3):          # in Python 3, range does xrange's job
    print(i, end=" ")
print()
print(type(range(3)))

Output:

NameError: name 'xrange' is not defined
0 1 2 
<class 'range'>

For more on loops over ranges, see for i in range in Python, and for counting down, reversing a range.

Memory: range stays tiny, lists don’t

A range only stores start, stop and step, so its size never changes. A list stores every number:

import sys

for n in (10, 1_000, 10_000_000):
    r = range(n)
    print(f"n = {n:>10,}: range {sys.getsizeof(r)} bytes | list {sys.getsizeof(list(r)):>11,} bytes")

Output:

n =         10: range 48 bytes | list         136 bytes
n =      1,000: range 48 bytes | list       8,056 bytes
n = 10,000,000: range 48 bytes | list  80,000,056 bytes

That’s exactly the saving xrange gave you in Python 2, and it’s why looping over range(10_000_000) doesn’t eat your RAM.

What a Python 3 range can do

Python 3’s range is more than a lazy counter. It supports len(), indexing, slicing, in, index() and count(), all computed with arithmetic instead of stored values:

r = range(0, 1_000_000_000, 5)     # a billion-wide range, created instantly

print(len(r))                      # 200,000,000 numbers, computed not stored
print(r[3], r[-1])                 # indexing works
print(r[10:15])                    # slicing gives another range
print(999_999_995 in r)            # membership is O(1) for integers
print(r.index(500), r.count(7))
print(list(reversed(range(5))))
print(range(0, 10, 2) == range(0, 9, 2))   # compared by the numbers they produce

Output:

200000000
15 999999995
range(50, 75, 5)
True
100 0
[4, 3, 2, 1, 0]
True

Is checking x in range fast?

For integers, yes: in on a range is a quick calculation, while in on a list scans item by item:

import timeit

n = 10_000_000
t_range = timeit.timeit(lambda: (n - 1) in range(n), number=1_000)
numbers = list(range(n))                              # build the list once
t_list = timeit.timeit(lambda: (n - 1) in numbers, number=20) / 20 * 1_000
print(f"x in range(n): {t_range * 1000:.3f} ms for 1,000 checks")
print(f"x in list:     {t_list:.0f} s for 1,000 checks (from 20 timed)")

Output:

x in range(n): 0.113 ms for 1,000 checks
x in list:     59 s for 1,000 checks (from 20 timed)

So don’t convert a range to a list just to test membership. For more on membership tests, see what in means in Python.

Porting xrange code to Python 3

The fix is almost always a plain search-and-replace of xrange( with range(. If one file has to run on both versions, alias it once:

try:
    xrange                       # Python 2
except NameError:
    xrange = range               # Python 3: alias it

total = 0
for i in xrange(1, 101):
    total += i
print("sum 1..100 =", total)
print("list when you need one:", list(range(5)))

Output:

sum 1..100 = 5050
list when you need one: [0, 1, 2, 3, 4]

The old 2to3 tool that automated this was removed in Python 3.13, so do it by hand or with a modern linter. Python 3 vs Python 2 lists the other big differences.

Related guides on loops and Python versions:

• Python for loop
• What does enumerate do in Python
• Install multiple Python versions
• len() function in Python

Frequently asked questions