The protection of cars against burglary and car theft is a major issue for automotive industry. Several techniques exist to stop the thief from driving away or to alert the owner. In this contribution we present an acoustic event detection (AED) framework for surveillance in cars, operating on a continuous audio stream which may be acquired by the existing in-car hands-free microphone. Furthermore we derive a new time-dependent accuracy measure tACC, evaluating both the accuracy and the temporal preciseness of the instant of detection. Operating on continuous audio, we examine a wide range of signal to noise ratios (SNRs) in a realistic parking scenario, showing the effectiveness of both the proposed AED framework and the new time-dependent accuracy measure.