For decades, vehicle emission regulations have been a cornerstone of air quality and climate policy1. They have helped cut smog, improve urban health, and stimulate innovation: leading to catalytic converters, cleaner combustion engines, and the growth of electric vehicles2. Regulatory frameworks now exist across all major markets, including the EU, U.S., China, India, and Japan3. But the scientific foundations of these regulations are still mostly laboratory tests.

Historically, type-approval procedures have relied on standardized driving cycles performed under controlled conditions (Fig. 1). These tests, such as the New European Driving Cycle (NEDC) and its successor, the Worldwide Harmonized Light Vehicles Test Procedure (WLTP), provide reproducible emissions and fuel consumption figures4. Yet they fall short in capturing how vehicles are actually used5,6,7. Test cycles typically assume mild acceleration, flat terrain, and consistent ambient conditions, assumptions that rarely match the variability of real-world driving.

Fig. 1: Summary of regulations evolution.
figure 1

In Europe, plug-in hybrid electric vehicles were regulated based on laboratory tests until the Diesel scandal in 2015. Since then, real-world data have been integrated and fully performance-based regulation is proposed for the future.

Full size image

As a result, type-approval values systematically underestimate actual emissions. In Europe, real-world CO₂ emissions of combustion engine vehicles are on average 40% higher than New-European-Driving-Cycle values, 20% higher than those according to the Worldwide Harmonized Light Vehicles Test Procedure, and even 300–500% higher for plug-in hybrid electric vehicles. These gaps distort vehicle design and supply, undermine policy credibility and delay climate progress8,9.

Regulators now have better tools and better data than ever before, and reliance solely on laboratory tests is no longer acceptable. A transition to real-world regulation will be complex but also transformative. We argue that a fundamental shift is both feasible and necessary: regulation of laboratory performance must be turned into regulation of real-world emissions.

Better data are available

For the first time in emission regulation history, policymakers are beginning to have the data infrastructure to regulate real-world emissions. Since 2021, the European Union has mandated all new cars to record real-world fuel consumption via On-Board Fuel Consumption Monitoring systems. More than 6.5 million vehicles are now reporting annual data, enabling comprehensive insights into fleet-level emissions under actual usage patterns10. In California, the Bureau of Automotive Repair has been collecting data via the On-Board Diagnostic that monitors emission control system components since 200011. These and similar sources are used primarily to update regulatory requirements, but they also create the foundation for a regulatory paradigm shift towards observed performance, instead of idealized lab conditions.

This shift is not without precedent. In response to the diesel emissions scandal, European regulators introduced Real Driving Emissions (RDE) tests for nitrogen oxides (NOₓ), using mobile measurement devices attached to cars during on-road trips12. The Real Driving Emissions protocol revealed just how far off laboratory figures had been and forced manufacturers to cut real-world NOₓ emissions from diesel vehicles significantly. The integration of RDE into formal compliance procedures demonstrated that real-world regulation works and leads to measurable air quality improvements.

Towards performance-based regulation

In fact, regulation of real-world vehicle emissions would be the natural next step towards applying performance-based regulations to vehicle emissions. Performance-based regulation itself is a long-standing concept in environmental policy regulating utilities, buildings, factory emissions but also nuclear and traffic safety13. Performance-based regulation focuses on achieving measurable, real-world outcomes rather than prescribing specific processes or test procedures14. It allows innovative designs that meet standards without mandating specific methods.

The core principle is simple: define clear performance targets and allow regulated entities flexibility in how they achieve them. This approach encourages innovation, improves cost-effectiveness, and focuses oversight on demonstrable results rather than procedural compliance. However, it also demands robust measurement systems, transparent data governance, and sophisticated monitoring capabilities to verify that intended outcomes are indeed achieved.

The case of plug-in hybrid electric vehicles

Plug-in hybrid electric vehicles are a compelling case for why performance-based policies that are based on real-world data are urgently needed. These vehicles combine an electric drivetrain with a combustion engine and rely heavily on usage behavior. Laboratory cycles assume daily charging and short trips, which results in low official CO₂ values. But most drivers do not charge daily, and many travel longer distances. As a result, average real-world fuel consumption is several times higher than reported through the Worldwide Harmonized Light Vehicles Test Procedure15. This discrepancy creates a policy blind spot: vehicles often appear cleaner and more efficient in regulation than they really are. Data gathering and analysis by researchers and non-governmental organizations has been slow, limited by budget and data sharing regulations. These efforts have produced useful updates but rarely led to large-scale action.

This is now changing. The European Commission is using on-board fuel consumption monitoring data in 2025 to adjust how plug-in hybrid electric vehicles count toward manufacturers’ fleet CO₂ targets16. Real-world usage patterns collected from the entire fleet of plug-in hybrid electric vehicles sold, such as electric driving share, will be used to weight each vehicle’s emissions contribution.

The inclusion of real-world data marks a historic move in emission regulation: observed data from everyday driving will directly influence regulatory compliance in a major emissions policy framework. A similar effort to update policies based on on-road performance was proposed by the California government, using data collected from on-board systems of plug-in hybrid electric vehicles in California during routine emission tests (SMOG test)17.

Real-world monitoring for transparency

Plug-in hybrid electric vehicles are just the beginning, similar methods could be applied to assess the performance of other vehicle types, including conventional hybrids, advanced combustion vehicles, and future low-emission technologies (cf. Fig. 1 for a summary of the historical evolution of vehicle emissions testing and the proposed shift towards performance-based emission regulation.

Laboratory tests still serve a role for type-approval of new technologies and component certification, but they must be supplemented by real-world surveillance. Performance-based regulation can close the gap between official targets and actual emissions, making policy more transparent, adaptive, and resilient to manipulation. It also aligns incentives for manufacturers, who must ensure that vehicles perform well not only under test conditions but throughout their lifecycle.

Messier metrics, better incentives

This transition raises both challenges and opportunities, as it involves shifting from controlled tests of technology to real-world metrics that depend on the interaction between the technology and user behavior. The performance of each vehicle manufacturer’s fleet will not be uniform based on one driving cycle and uniform charging behavior, but on the consumers of the specific vehicle fleet, which introduces uncertainty for automakers, who may face variable compliance outcomes depending on vehicle usage.

The transition to real-world measurements also creates new opportunities for automakers to improve performance through several mechanisms, also directed at user behavior such as eco-feedback, vehicle tuning, enhanced charging opportunities, vehicle maintenance, tires and roof rack selections and other strategies that influence on-road vehicle performance but are not captured by current measurement tools. Furthermore, policy based on real-world data may help sustain performance over time.

Expand, align, invest

Implementing such a policy raises questions about data access, privacy, and governance. Regulators must ensure that data are anonymized, robustly audited, and fairly applied. Yet these are solvable problems. If cities can manage real-world air pollution budgets and citizens are held accountable for vehicle inspections and roadworthiness, it is not unreasonable to hold manufacturers accountable for real-world fleet emissions.

We recommend that vehicle emission regulations evolve in three directions. First, expand the use of real-world data for compliance, as is already happening with plug-in hybrid electric vehicles. The strategies used can be extended to other vehicle types and segments. Second, harmonize data collection standards internationally to enable comparisons and alignment across regions. And third, invest in research to design fair, effective performance-based frameworks that consider variability and behavioral factors. These frameworks should become sufficiently flexible to adjust regulatory measures in response to observed, real-world performance rather than relying on predetermined outcomes.

The performance-based regulation of plug-in hybrid electric vehicles in Europe can serve as a model for broader policy innovation across transportation, energy, and environmental and additional sectors. For the sake of public trust, policy integrity, and climate progress, it is time to regulate what happens on the road.