For many years Distributed Denial-of-Service attacks have been known to be a threat to Internet services. Recently a configuration flaw in NTP daemons led to attacks with traffic rates of several hundred Gbit/s. For those attacks a third party, the amplifier, is used to significantly increase the volume of traffic reflected to the victim. Recent research revealed more UDP-based protocols that are vulnerable to amplification attacks. Detecting such attacks from an abused amplifier network’s point of view has only rarely been investigated.In this work we identify novel properties which characterize amplification attacks and allow to identify the illegitimate use of arbitrary services.Their suitability for amplification attack detection is evaluated in large high-speed research networks. We prove that our approach is fully capable of detecting attacks that were already seen in the wild as well as capable of detecting attacks we conducted ourselves exploiting newly discovered vulnerabilities.