It is postulated that feedforward artificial neural networks can be used for fast and robust parametric testing of mixed-signal circuits when applied to the processing of transient waveforms which are circuit responses to test signals. Numerical and experimental results are presented to verify the validity of the technique using examples of OPAMP and OTA-C filters and of a CMOS inverter. A feedforward artificial neural network in the form of a single-hidden-layer sigmoidal perceptron is trained in this preliminary study to estimate the circuit parameters. Since no iterative calculations are performed to identify parameter values with this technique, it is highly suitable for high-speed parametric testing. It is also more robust in the presence of noise when compared to traditional approaches. Topics for future research are addressed.