Our Methodology
Data-driven insights into the real-world reliability and major repair costs of vehicles from 2010 to the present.
Our platform provides data-driven insights into the real-world reliability and major repair costs of vehicles. To ensure our data is accurate, actionable, and reflective of actual ownership experiences, we have developed a rigorous methodology for gathering, structuring, and scoring vehicle issues.
Data Collection and Sources
Our database focuses exclusively on major repair issues—defined as common, well-documented failures that typically cost $1,000 or more to repair. We do not track routine maintenance (e.g., brake pads, oil changes) or minor cosmetic defects unless they lead to significant secondary damage.
To identify these issues, we aggregate data from multiple authoritative and community-driven sources:
- NHTSA Complaint Data: We analyze complaint volumes submitted to the National Highway Traffic Safety Administration (NHTSA) to identify widespread safety and mechanical defects.
- Technical Service Bulletins (TSBs) and Recalls: Manufacturer-issued TSBs provide concrete evidence of known design flaws and the official repair procedures required to address them.
- Enthusiast Forums and Owner Communities: Model-specific forums (e.g., Bimmerpost, Rennlist, Toyota Nation) are invaluable for identifying emerging issues before they reach the level of a formal recall, as well as establishing real-world repair cost ranges.
- Independent Repair Data: We cross-reference repair cost estimates with data from independent mechanics and specialized repair shops to ensure our cost ranges reflect actual out-of-pocket expenses for owners.
Engine-Specific Tracking
A common flaw in many reliability ratings is grouping an entire vehicle model under a single score. For example, a BMW 3 Series with a reliable B58 engine will have a vastly different ownership experience than the same model equipped with an early N20 engine.
To solve this, our database tracks issues at the Engine-Specific Variant level. Each vehicle is assigned a unique identifier that combines the make, model, generation, and specific engine variant (e.g., lexus-is-xe30-2.0l-turbo vs.lexus-is-xe30-3.5l-v6).
When an issue is identified, it is classified by scope:
- Engine-Specific: Issues that only affect a particular powertrain (e.g., turbocharger failure) are assigned only to that engine variant.
- Shared: Issues related to chassis, suspension, transmission, or electronics that affect all variants are duplicated across all applicable engine slugs.
- Hybrid/EV-Specific: Issues related to high-voltage battery degradation or inverter failures are assigned exclusively to hybrid or PHEV variants.
The 5-Tier Severity Scale
Repair cost alone does not tell the whole story. A $1,500 head gasket failure that destroys the engine if ignored is fundamentally more severe than a $2,000 infotainment screen replacement. To provide a more accurate picture of risk, we developed a 5-Tier Severity Scale that weights the financial cost of the repair equally with the consequence of ignoring the issue.
| Tier | Label | Cost Range | Definition | Examples |
|---|---|---|---|---|
| 1 | Low | $1,000 – $1,800 | Inconvenient but not urgent; carries no risk of further mechanical damage if deferred. | Oil filter housing leaks, minor sensor failures |
| 2 | Moderate | $1,800 – $3,500 | Should be addressed soon; may cause secondary damage or drivability issues if ignored. | Water pump failure, boost pipe leaks |
| 3 | High | $3,500 – $6,000 | Significant repair; causes further damage or immediate safety concerns if deferred. | Timing chain stretch, head gasket failure |
| 4 | Severe | $6,000 – $12,000 | Major mechanical repair; vehicle may become undriveable; carries a risk of total loss. | Rod bearing failure, DCT mechatronic failure |
| 5 | Extreme | $12,000+ | Catastrophic failure; repair cost often approaches or exceeds the total value of the vehicle. | Full engine replacement, EV battery pack replacement |
The Prevalence Tier
A critical flaw in traditional reliability reporting is conflating the existence of an issue with its frequency. A catastrophic engine failure might be well-documented, but if it only affects 1% of owners, it should not carry the same weight as a transmission failure that affects 40% of owners.
To solve this, we separate our confidence in the data from the actual frequency of the issue using a Prevalence Tier. This metric indicates how commonly an issue occurs in the real world, allowing users to distinguish between rare anomalies and chronic design flaws.
| Prevalence Tier | Meaning | Real-World Implication |
|---|---|---|
| ◆◆◆◆◆ Very Common | Chronic, widespread issue | Likely to affect a significant percentage of owners; buyers should expect to encounter this. |
| ◆◆◆◆ Common | Frequently reported | A known weak point of the vehicle; buyers should check if it has been addressed. |
| ◆◆◆ Uncommon | Occasional reports | Happens often enough to be documented, but not a guarantee for every owner. |
| ◆◆ Rare | Isolated incidents | The issue is real and documented, but affects a very small minority of vehicles. |
| ◆ Very Rare | Extreme outliers | Highly unusual failures, often related to severe neglect or extreme edge cases. |
Objective Backing: Where applicable, the qualitative Prevalence Tier is backed by an objective NHTSA Complaint Count. This provides a concrete, citable number of reported incidents for a specific issue, adding a layer of verifiable data to our prevalence ratings.
Confidence Scoring
Our Confidence Score (High, Medium, Low) does not reflect how often an issue happens—that is handled by the Prevalence Tier. Instead, the Confidence Score reflects the quality and volume of our research sources.
- High Confidence: The issue is corroborated by multiple independent sources, such as official TSBs, high NHTSA complaint volumes, and widespread consensus on owner forums.
- Medium Confidence: The issue is reported by owners and independent mechanics, but may lack an official TSB or widespread media coverage.
- Low Confidence: The issue is based on isolated reports or anecdotal evidence that requires further verification.
The Reliability Score Algorithm
Our A-F reliability grade is calculated using a weighted scoring algorithm that starts at 100 points and applies penalties based on multiple factors. The final score is converted to a letter grade (A: 85-100, B: 70-84, C: 55-69, D: 40-54, F: 0-39).
Penalty Factors
1. NHTSA Recall Penalty
Recalls ≤ 3: 5 points each | Recalls > 3: Logarithmic scaling with diminishing returns
2. NHTSA Complaint Penalty
Normalized per 10,000 units sold × 2 (capped at 20 points)
3. Issue Severity × Prevalence Penalty
Each issue contributes: Confidence × Severity Weight × Prevalence Multiplier
- Low (1): 0.8×
- Moderate (2): 2×
- High (3): 4×
- Severe (4): 7×
- Extreme (5): 10×
- Very Rare: 0.2×
- Rare: 0.4×
- Uncommon: 1×
- Common: 1.2×
- Very Common: 1.4×
4. Repair Cost Adjustment
Average repair cost < $800: 0.85× penalty (reward) | $800-$1,500: 0.85-1.0× | $1,500-$2,500: 1.0-1.1× | > $2,500: 1.1-1.2× (penalty increases with cost)
By separating Severity, Prevalence, Confidence, and Cost, our methodology provides a nuanced, highly accurate assessment of vehicle reliability that empowers buyers to make informed decisions.
Ready to Find Your Next Vehicle?
Browse our reliability rankings to compare vehicles by score, budget, and category.
View Rankings