A computer-implemented method is provided for screening communication traffic. A list of one or more threshold pairs including respective first and second threshold values, each of which first threshold values is greater than one, are defined. Network traffic from a plurality of sources is monitored, so as to determine for each source a count of unique destination addresses to which the source transmitted data during a period of time. A response to malicious network traffic is invoked responsively to finding that each of a first number of the sources sent data to at least a second number of the destination addresses, wherein, for at least one of the threshold pairs, the first number is at least equal to the respective first threshold value, and the second number is at least equal to the respective second threshold value.