kfir,
Although you can't partition a table based on column values, you might look into the RAID syntax of the CREATE TABLE statement. It's primarily used for manually striping the data across multiple files, instead of placing all data instead of a single file. Sort of table-level round-robin partitioning/striping...
Another option I use is to physically partition the table based on a primary key. I've broken up a single table into ten separate tables that have the same schema. When an incoming sql request comes in the integer primary key value used in the sql statement is evaluated using modulus. If the remainder of the modulus of 10 for example is 4, go to table 4 for the data.