Pandas version checks

  • [x] I have checked that this issue has not already been reported.

  • [x] I have confirmed this bug exists on the latest version of pandas.

  • [x] I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

import pandas as pd

_s = pd.Series([    {"k": f"{i}", "m": "q"} for i in range(5)
])

pd.json_normalize(_s,record_prefix="T")

Issue Description

The record_prefix is completely ignored and the resulting columns are note renamed.

   k  m
0  0  q
1  1  q
2  2  q
3  3  q
4  4  q

The issue is in L517-L527 where a short cut is taken for speed. One option to fix this would be to pass the record_prefix as prefix to nested_to_record.

This is hidden in the documentation example by using a positional argument for the record path

    >>> data = {"A": [1, 2]}
    >>> pd.json_normalize(data, "A", record_prefix="Prefix.")
        Prefix.0
    0          1
    1          2

Expected Behavior

The prefix should be applied (or an explicit warning be given)

  T.k T.m
0   0   q
1   1   q
2   2   q
3   3   q
4   4   q

Installed Versions

INSTALLED VERSIONS ------------------ commit : c888af6d0bb674932007623c0867e1fbd4bdc2c6 python : 3.13.3 python-bits : 64 OS : Linux OS-release : 6.14.0-28-generic Version : #28-Ubuntu SMP PREEMPT_DYNAMIC Wed Jul 23 12:05:14 UTC 2025 machine : x86_64 processor : x86_64 byteorder : little LC_ALL : None LANG : en_US.UTF-8 LOCALE : en_US.UTF-8 pandas : 2.3.1 numpy : 2.3.2 pytz : 2025.2 dateutil : 2.9.0.post0 pip : 25.0 Cython : None sphinx : None IPython : 9.4.0 adbc-driver-postgresql: None adbc-driver-sqlite : None bs4 : 4.13.4 blosc : None bottleneck : 1.5.0 dataframe-api-compat : None fastparquet : 2024.11.0 fsspec : 2025.5.1 html5lib : 1.1 hypothesis : 6.135.20 gcsfs : 2025.5.1 jinja2 : 3.1.6 lxml.etree : 5.4.0 matplotlib : 3.10.5 numba : None numexpr : 2.11.0 odfpy : None openpyxl : 3.1.5 pandas_gbq : 0.29.1 psycopg2 : None pymysql : 1.4.6 pyarrow : 21.0.0 pyreadstat : 1.3.0 pytest : 8.4.1 python-calamine : None pyxlsb : 1.0.10 s3fs : 2025.5.1 scipy : 1.16.1 sqlalchemy : 2.0.43 tables : 3.10.2 tabulate : 0.9.0 xarray : 2025.6.1 xlrd : 2.0.2 xlsxwriter : 3.2.5 zstandard : 0.23.0 tzdata : 2025.2 qtpy : 2.4.3 pyqt5 : None

Comment From: SumitkCodes

@JensHeinrich Can you check #62215

Comment From: JensHeinrich

@JensHeinrich Can you check #62215

A quick check seems to show it working as expected now. Thanks!

Comment From: rhshadrach

Thanks for the report. From the documentation, best I can guess, the intention was for record_prefix to only apply when record_path was specified. However, it does make sense to me to apply this prefix regardless. The code that applies this prefix later in the method can be used here too:

if record_prefix is not None:
    result = result.rename(columns=lambda x: f"{record_prefix}{x}")

I'm positive on this - PRs welcome!

Comment From: JensHeinrich

Thanks for the report. From the documentation, best I can guess, the intention was for record_prefix to only apply when record_path was specified. However, it does make sense to me to apply this prefix regardless. The code that applies this prefix later in the method can be used here too:

if record_prefix is not None: result = result.rename(columns=lambda x: f"{record_prefix}{x}")

My instinct would have been similar to #62215 and change https://github.com/SumitkCodes/pandas/blob/3e1d6d5bc853bd8bc983291b12aec2dbf477dde6/pandas/io/json/_normalize.py#L526

if record_prefix:
    data = nested_to_record(data, prefix=record_prefix, sep=sep, max_level=max_level + 1, level = 1)
else:
    data = nested_to_record(data, sep=sep, max_level=max_level)

But i just realized this would behave different from record_path-path as it would be 

 result = result.rename(columns=lambda x: f"{record_prefix}{sep}{x}")

instead of

 result = result.rename(columns=lambda x: f"{record_prefix}{x}")

But ignoring the sep at https://github.com/SumitkCodes/pandas/blob/3e1d6d5bc853bd8bc983291b12aec2dbf477dde6/pandas/io/json/_normalize.py#L580 is unexpected IMO

Comment From: PluvioXO

Done a solution that should be optimal and passes made unit tests sending PR now.

Comment From: PluvioXO

From what I can see that should fix it, if I missed something lmk @JensHeinrich