PHP uses a blend of interpretation and compilation in order to provide the best mix of performance and flexibility to programmers.
Behind the scenes, PHP compiles your script down to a series of instructions (called opcodes) whenever it is accessed. These instructions are then executed one by one until the script terminates.
This constant recompilation may seem a waste of processor time, but it is actually not all that bad because you no longer need to worry about hand recompiling your scripts when you make any changes.
Furthermore, it provides very quick feedback during development. If you have an error somewhere in your file, PHP will refuse to compile the page until you have fixed the problem, and you are able to step through execution of your code line by line until you find the problem.
The speed hit of regular compilation is nullified entirely by the use of PHP opcode caches.
One major advantage to having interpreted code is that all memory used by the script is managed by PHP, and the language automatically cleans up after every script has finished. This means that you do not need to worry about closing database links, freeing memory assigned to images, and so on, because PHP will do it for you. That is not to say you should be lazy and make PHP do all the work – good programmers clean up themselves, and let PHP work as backup in case something is missed.