PyInns - Home

Iterating with .iterrows() in pandas – Why You Should Avoid It in 2026

df.iterrows() is one of the most commonly used — and most criticized — methods in pandas. While it looks convenient, it is notoriously slow and should be avoided in almost all cases in 2026.

This March 15, 2026 guide explains why .iterrows() is slow and shows the modern, efficient alternatives you should use instead.

TL;DR — Key Takeaways 2026

.iterrows() is very slow because it returns a Series for each row
It is one of the worst ways to iterate over a DataFrame
Prefer vectorized operations, .itertuples(), or .apply() (sparingly)
Vectorized code is typically 50–200x faster than .iterrows()
Never use .iterrows() in production code unless absolutely necessary

1. The Problem with .iterrows()


import pandas as pd

df = pd.DataFrame({
    'name': ['Alice', 'Bob', 'Charlie'],
    'score': [85, 92, 78]
})

# ❌ Very slow and not Pythonic
for index, row in df.iterrows():
    df.loc[index, 'score_double'] = row['score'] * 2

2. Better Alternatives in 2026


# ✅ Option 1: Vectorized (Recommended)
df['score_double'] = df['score'] * 2

# ✅ Option 2: itertuples() - Much faster than iterrows
for row in df.itertuples():
    # row is a namedtuple - very fast access
    pass

# ✅ Option 3: When you really need row-wise logic
df['score_double'] = df.apply(lambda row: row['score'] * 2, axis=1)

3. Performance Comparison 2026

Method	Relative Speed	Recommendation
Vectorized operations	100x (fastest)	Best choice
`.itertuples()`	20–50x	Acceptable when needed
`.apply(axis=1)`	5–15x	Use sparingly
`.iterrows()`	1x (slowest)	Avoid in production

4. Best Practices in 2026

Never use .iterrows() for performance-critical code
Default to vectorized operations whenever possible
Use .itertuples() when you must iterate row by row
Use .apply() only when vectorization is not feasible
Profile first — then decide if iteration is truly needed

Conclusion — Iterating with .iterrows() in 2026

.iterrows() is one of the slowest ways to iterate over a pandas DataFrame. In 2026, the most efficient pandas code almost never uses .iterrows(). Instead, it relies on vectorized operations, boolean indexing, and occasionally .itertuples() when row-by-row processing is unavoidable.

Next steps: