PrintStream was the original bridge to deal with encoding characters and other datatypes. If you look at the javadoc for java.io.OutputStream you’ll see methods only for writing two distinct data types: byte and int.
In early versions of the JDK (1.0.x), when you wanted to write characters, you could do one of two things, write bytes to an output stream (which are assumed to be in the system default character set):
outputStream.write("foobar".getBytes());
or wrap another outputStream in a PrintStream:
PrintStream printStream = new PrintStream(outputStream);
printStream.write("foobar");
See the difference? PrintStream is handling the character conversion to bytes, as well as encoding (the constructor call above uses the system default encoding, but you could pass it as a parameter). It also provides convenience methods for writing double, boolean, etc….
In fact System.out and System.err are defined as PrintStream instances.
Along comes JDK 1.1, and they realize they need a better way to deal with pure character data, since PrintStream still has the byte based methods for writing. So they introduced the Writer abstract class to deal strictly with char, String and int data.
PrintWriter adds methods for other types like double, boolean, etc…
Nowadays PrintWriter also has format() / printf() methods for format printing, etc…
As a general rule, if you’re writing character data, use Writer instances. If you’re writing binary (or mixed) data use OutputStream instances.