To give you some very general advice, always use original sources where possible. Where you haven't been able to locate the original source (or when it's not really necessary as it's not a key reference for your project), make sure you make the reader know that you are referencing a secondary source. e.g. if you read in a review (written by Smith et al, 2007) that Jones et al, 2005 report blah, then you would cite it as
blah (Jones et al, 2005 cited by Smith et al, 2007)
As for what you need to reference...well, if in doubt, reference. If it's a very clear and obvious statement known to all likely to be reading your work, then you don't need to reference.
Good luck!!