Most everyday Python code never touches individual bits directly — but python bitwise operators show up more often than you'd expect, from permission flags to performance tricks to how Python's own set operations are named. This article covers all six, including why one of them produces results that look confusingly negative.
What Are Bitwise Operators?
Bitwise operators work directly on the individual bits of an integer's binary representation, rather than on the number as a whole. This is fundamentally different from the logical operators covered in an earlier article — and, or, and not work with entire Boolean truth values (True/False), while bitwise operators work bit by bit, on the underlying binary digits that make up a number.
Python has six bitwise operators:
& # AND
| # OR
^ # XOR
~ # NOT
<< # left shift
>> # right shiftTo make sense of any of these, it helps to remember that every integer is stored internally as a sequence of binary digits (bits) — 5, for instance, is 101 in binary. Bitwise operators compare or shift those digits directly.
Bitwise AND, OR, and XOR
& (AND): 1 only if both bits are 1
print(5 & 3) # 1Working through it in binary:
5 = 101
3 = 011
---
001 (only the rightmost bit is 1 in both)& compares each corresponding pair of bits — the result bit is 1 only where both inputs have a 1. This makes & useful for masking: isolating specific bits you care about while zeroing out the rest.
| (OR): 1 if either bit is 1
print(5 | 3) # 75 = 101
3 = 011
---
111 (a bit is 1 if it's 1 in either number)| is commonly used to combine flags — setting multiple binary options at once without disturbing bits that are already set.
^ (XOR): 1 if exactly one bit is 1
print(5 ^ 3) # 65 = 101
3 = 011
---
110 (1 only where the bits differ)XOR (exclusive or) is the one people find least intuitive at first. It returns 1 only when the two bits differ — if both bits are the same (both 0 or both 1), the result is 0. This gives XOR some genuinely useful properties: it's the basis for toggling a bit on and off with the same operation, for simple (non-cryptographically-secure) encryption schemes, and for quickly finding what's different between two values.
Bitwise NOT (~)
~ inverts every bit in a number — every 0 becomes a 1, and every 1 becomes a 0. This is sometimes called the "one's complement."
print(~5) # -6Why the result looks negative
This is the part that confuses almost everyone the first time they see it. Python (like most languages) represents integers using a system called two's complement, where the leftmost bit effectively signals whether a number is negative. Flipping every bit of a positive number necessarily flips that sign bit too, which is why ~5 doesn't give you some large unsigned binary value — it gives you a negative number instead.
There's a simple formula that captures this behavior without needing to think through the binary representation each time:
# ~x is always equivalent to -x - 1
print(~5) # -6 → -(5) - 1
print(~0) # -1 → -(0) - 1
print(~-1) # 0 → -(-1) - 1That formula — ~x == -x - 1 — is worth just memorizing, since it explains the output far faster than mentally flipping bits.
Left and Right Shift Operators (<<, >>)
Left shift (<<)
<< shifts a number's bits to the left by a given number of positions, filling in zeros on the right. Each shift to the left is mathematically equivalent to multiplying by 2, raised to the power of however many positions you shifted:
print(4 << 1) # 8 — 4 * (2 ** 1)
print(4 << 2) # 16 — 4 * (2 ** 2)
print(1 << 3) # 8 — 1 * (2 ** 3)Right shift (>>)
>> does the reverse — it shifts bits to the right, which is equivalent to floor-dividing by 2 raised to the power of the shift amount:
print(16 >> 1) # 8 — 16 // (2 ** 1)
print(16 >> 2) # 4 — 16 // (2 ** 2)
print(9 >> 1) # 4 — 9 // 2, floor division discards the remainderPractical example: fast multiplication and division by powers of two
Because these operators map directly to multiplying or dividing by powers of two, they're occasionally used as a performance shortcut in code where every cycle matters — particularly in lower-level or performance-sensitive contexts:
value = 10
doubled = value << 1 # equivalent to value * 2
halved = value >> 1 # equivalent to value // 2
print(doubled, halved) # 20 5In typical, everyday Python code, this micro-optimization rarely matters — write value * 2 for clarity unless you have a specific, measured reason to reach for the shift operator instead.
Real-World Uses and Best Practices
Common applications
Bitwise operators show up in a handful of recognizable situations:
Bit masking — isolating or clearing specific bits within a larger value, common in low-level data parsing.
Flags and permissions — representing several independent on/off options as a single integer, where each bit represents one flag. This pattern appears in file permission systems and various configuration formats.
Low-level system programming — networking code, hardware interfaces, and file format parsing frequently need to manipulate individual bits directly.
Simple XOR-based encryption — XOR's "toggle" property makes it a building block in basic obfuscation or encryption schemes, though modern cryptography relies on far more sophisticated algorithms for anything security-sensitive.
Using bin() to visualize binary
While you're learning, Python's built-in bin() function is genuinely useful for seeing exactly what's happening at the bit level, rather than working it out by hand:
print(bin(5)) # 0b101
print(bin(3)) # 0b11
print(bin(5 & 3)) # 0b1
print(bin(5 | 3)) # 0b111Running your bitwise expressions through bin() while you're first getting comfortable with these operators makes the binary math tangible instead of abstract.
Operator precedence
Bitwise operators bind lower than most arithmetic operators in Python — meaning arithmetic gets evaluated before bitwise operations, unless you specify otherwise. This can produce results that don't match what you'd naturally assume:
result = 1 << 2 + 3
print(result) # 32 — not what you might expect
# What actually happened: addition ran first, then the shift
# 1 << (2 + 3) → 1 << 5 → 32To avoid this kind of surprise, wrap bitwise operations in explicit parentheses any time they're mixed with arithmetic:
result = (1 << 2) + 3
print(result) # 7 — the shift happens first, as intendedA note on operator overloading
Python allows custom classes to define their own behavior for these operators through special "dunder" methods — __and__() for &, __or__() for |, __xor__() for ^, and so on. This is a more advanced topic you'll encounter later when working with classes, but there's a concrete example of it you've likely already used without realizing it: Python's built-in set type reuses &, |, and ^ for set operations — intersection, union, and symmetric difference, respectively:
a = {1, 2, 3}
b = {2, 3, 4}
print(a & b) # {2, 3} — intersection
print(a | b) # {1, 2, 3, 4} — union
print(a ^ b) # {1, 4} — symmetric difference (in one set or the other, not both)This isn't a coincidence — it's a deliberate example of operator overloading in Python's own standard library, showing how the same symbols can mean something entirely different depending on the type of the operands involved.