If you have many calculations that do not depend on each-others solutions (aka embarrassingly parallel - e.g. Monte Carlo) you can use spmd function with great success.
They provide a lot of useful information about parallel processing and speeding up your computation using Matlab. Having watched these, you should be able to handle parallel processing quite well.
To add, if run time increases from using parallel processing, it means that the time spent to handle the overhead required to set up parallel processes is greater than the time saved by using parallel processes. If this is the case parallel processing probably isn't necessary anyway.