To meet US EPA light-duty vehicle emission standards, the vehicle powertrain has to be optimally controlled in addition to maintaining very high catalyst system efficiency. If vehicles are operated outside the bounds of a standard laboratory exhaust emission test (e.g., on-road or off-cycle) the operating control strategy may shift to optimize other desirable parameters such as fuel economy and drivability. Under these circumstances. The engine control system could be operating in a different state space from an emission control stand point. This control state-space can be observed based on four principal parameters: NOx, Lambda and exhaust temperature (measured at the tailpipe) and vehicle acceleration. These vehicle emission control patterns can be characterized by their corresponding emission control signatures, such as cold start, transient fuel control, and high speed/high load open loop. These emission control signatures are unique to a variety of engine technologies as well. Recognizing these signatures during vehicle operation can identify engine control state space and could estimate NOx mass flow by utilizing an ANN (artificial neural network) for pattern recognition. This could assist in detecting emission testing irregularities that might indicate a malfunctioning emission control system. One advantage to this approach is the equipment overhead to acquire this information is much less compared to other conventional methods such as PEMS (portable emission measurement system). US EPA is investigating this approach, recording the vehicle emission control dynamic signatures during normal dynamometer testing and on-road/off-cycle. Optimized data sets of emission control signatures are currently being used for training an artificial neural network to estimate NOx mass-based calculations and distinguish between well-controlled and uncontrolled systems. This non-intrusive testing method may be used to detect catalyst early failure and monitor emission test irregularities.