At the transport level, you often don't have many choices. Applications you use on your network will generally use TCP or UDP. Because TCP is connection oriented and reliable, it is generally preferable to UDP from a security perspective. Keep in mind that UDP is often faster because of the overhead TCP adds.
One of the most significant reasons TCP is more secure than UDP is the difficulty in spoofing TCP communications. As you learned in Chapter 3, UDP spoofing is trivial since there is no notion of connection. This is a main reason why UDP protocols such as SNMP, TFTP, and syslog need special attention when deployed in a security-sensitive environment. Spoofing TCP SYN packets is also easy because no response is needed by the host. (The connection hasn't been formed at this point.) Trying to hijack an established TCP session, however, is very difficult if the attacker is unable to see the packets flow on the wire. This is because the 32-bit sequence number must be guessed by the attacker. More details on UDP and TCP spoofing (including header diagrams) can be found in the "Spoof" section of Chapter 3.
In the past, initial sequence numbers (ISNs) were not sufficiently random. (In the Kevin Mitnick attack against Tsutomu Shimomura's computers, the ISN incremented by 128,000 for each new session.) Today, however, most modern systems choose much better ISNs that are difficult for an attacker to guess.