A sample of Carfax marketplace listings with vehicle specs, mileage, dealer ratings, history verification, reliability badges, and One Price value estimates.

This repository contains a preview sample of the Carfax dataset published by Rebrowser. If you're doing academic research, you may be eligible for free access to a much larger slice — see Free Datasets for Research.

This dataset contains 1 entity, each in its own folder: Car Listings (car-listings). See below for a full field breakdown, sample counts, and data distributions for each.

Found this useful? ⭐ Star this repo to help us keep publishing fresh data. Found an error? Let us know.

A sample of Carfax marketplace listings with year, make, model, mileage, body style, drivetrain, dealer location, ratings, history verification, and reliability data.

4,303,120 total records from 2025-11-16 to 2026-05-03, up to 30,000 rows in this sample (0.70% of full dataset). Exported as one file per day, up to 1,000 rows each, last 30 days retained.

🔒 Premium fields are included in the data files but their values are replaced with [PREMIUM]. To access real values, use our website.

Rebrowser web viewer lets you filter, sort, and export any slice of this dataset interactively. These pre-built views are ready to open:

Listings with History Verification — 3,958,277 records

↳ [{"sort":"currentPrice DESC"}]

Clean Title No-Accident Listings — 2,669,665 records

↳ [{"field":"noAccidents","op":"isTrue"},{"sort":"currentPrice DESC"}]

New Vehicle Listings — 572 records

↳ [{"field":"vehicleCondition","op":"is","value":"New"},{"sort":"currentPrice DESC"}]

Advantage Dealer Listings — 3,922,773 records

↳ [{"field":"advantage","op":"isTrue"},{"sort":"sellerRating DESC"}]

One-Owner Vehicle Listings — 1,993,362 records

↳ [{"field":"oneOwner","op":"isTrue"},{"sort":"mileage ASC"}]

See all 28 views →

import pandas as pd
from pathlib import Path

# ── Car Listings ───────────────────────────────────────────────────────────
# Load the last 7 days of listings
files = sorted(Path('rebrowser/carfax-dataset/car-listings/data').glob('*.parquet'))[-7:]
listings = pd.concat([pd.read_parquet(f) for f in files])

# Top 10 makes by listing count
print(listings['make'].value_counts().head(10).to_string())

# Average mileage by body style
print(listings.groupby('bodyStyle')['mileage'].mean().sort_values().to_string())

# No-accident listings by state — top 10 states
clean = listings[listings['noAccidents'] == True]
print(clean['sellerState'].value_counts().head(10).to_string())

# Dealer ratings: top-rated sellers with 50+ reviews
top_dealers = listings[listings['sellerReviewCount'] >= 50]
print(top_dealers.groupby('sellerId')['sellerRating']
      .first().sort_values(ascending=False).head(10).to_string())

# Badge distribution (GOOD, GREAT, FAIR) by drivetrain
print(pd.crosstab(listings['drivetrain'], listings['badge']).to_string())

# One-owner SUVs with service records
one_owner_suvs = listings[
    (listings['oneOwner'] == True) &
    (listings['bodyStyle'] == 'SUV') &
    (listings['serviceRecords'] == True)
]
print(f"One-owner SUVs with service records: {len(one_owner_suvs)}")
print(one_owner_suvs[['make', 'model', 'year', 'mileage', 'sellerState']]
      .sort_values('mileage').head(10).to_string(index=False))

Build regression models predicting used car values from mileage, body style, drivetrain, and market location. Validate against One Price estimates and reliability badges.

Rank dealers by rating, review count, and inventory size across states. Identify top-performing dealer groups and study how seller reputation correlates with listing activity.

Analyze body style, fuel type, and drivetrain distributions across geographic markets. Detect which vehicle segments are over- or under-represented in specific states.

Study how no-accident history, one-owner status, and service record counts vary by make, model, and age. Quantify the frequency of clean-history vehicles across market segments.

This repo is a 1,000-row preview sample. The full dataset is at rebrowser.net/products/datasets/carfax

Doing academic research? You may qualify for free access to a larger slice. See Free Datasets for Research.

On Rebrowser you can:

• Filter before you buy — use the web UI to apply filters on any field and sort by any column. Preview results before purchasing. You only pay for records that match your criteria.
• Export in your format — CSV, JSON, JSONL, or Parquet depending on your plan.
• Access via API — integrate dataset queries into your pipelines and workflows.
• Choose your freshness — plans range from a 14-day lag to real-time data with no delay.
• Select only the fields you need — keep exports lean. Premium fields with richer data are available on higher plans.

Pricing starts at $2 per 1,000 rows with volume discounts.

Free for research and non-commercial use with attribution. See license terms and how to cite.

@misc{rebrowser_carfax,
  author       = {Rebrowser},
  title        = {Carfax Used Car Listings & Dealer Inventory Dataset},
  year         = {2026},
  howpublished = {\url{https://rebrowser.net/products/datasets/carfax}},
  note         = {Accessed: YYYY-MM-DD}
}

Commercial use requires a paid license — see pricing. Use of this data is governed by the Rebrowser Terms of Use, which may be updated at any time independently of this repository.

Rebrowser is an independent data provider and is not affiliated with, endorsed by, or sponsored by Carfax. Any trademarks are the property of their respective owners. This dataset is compiled from publicly available information; we do not request or collect Carfax user credentials. By using this dataset, you agree to comply with Carfax's Terms of Service and all applicable laws and regulations. Images, logos, descriptions, and other materials included in this dataset remain the intellectual property of their respective owners and are provided solely for informational purposes. Rebrowser makes no warranties regarding the accuracy, completeness, or legality of the data and assumes no liability for how the data is used. You are solely responsible for ensuring that your use of this dataset does not infringe on the rights of any third party.

You can also find this data on Kaggle, HuggingFace, Zenodo.