One major problem with this technique is that, when a firewall blocks outgoing ICMP Port Unreachable messages, the port will appear open. These false-positives are hard to distinguish from real open ports.
Another disadvantage with UDP scanning is the speed at which it can be performed. Most operating systems limit the number of ICMP Port Unreachable messages which can be generated in a certain time period, thus slowing the speed of a UDP scan. Nmap adjusts its scan speed accordingly to avoid flooding a network with useless packets. An interesting point to note here is that Microsoft do not limit the Port Unreachable error generation frequency, and thus it is easy to scan a Windows machine's 65,535 UDP Ports in very little time!!
UDP Scanning is not usually useful for most types of attack, but it can reveal information about services or trojans which rely on UDP, for example SNMP, NFS, the Back Orifice trojan backdoor and many other exploitable services.
Most modern services utilise TCP, and thus UDP scanning is not usually included in a pre-attack information gathering exercise unless a TCP scan or other sources indicate that it would be worth the time taken to perform a UDP scan.