It really depends on the kind of experiment you're trying to do. Can you tell us a little bit more about what you're trying to measure?
Also, just to check - do you mean miRNA (as in, microRNA - the 21-24nt, non-coding RNAs) or mRNA (as in messenger RNA, the long RNAs that code for proteins)? Sorry if this is a silly question. miRNA are very small (hence microRNA!) and, given that your primer for cDNA synthesis is typically going to be as big as the miRNA itself, you woudn't be able to measure it by normal RT-PCR. While you can measure miRNA by RT-PCR, it's usually a fairly complicated process that involves ligating RNA linkers onto the 5' and 3' ends before amplification. mRNA is of course much easier to amplify and use in RT-PCR - as I said, it depends on what you're trying to do.
When I'm doing RT-PCR for a particular mRNA (e.g. trying to test if its transcription changes in a mutant genotype or in response to a treatment) I usually just purify total RNA and use an oligo(dT)20 primer for cDNA synthesis - the oligo(dT) will enrich for the polyadenylated RNA anyway, so you'll be mainly making cDNA from your mRNA anyway and consequently don't need to specifically purify mRNA (by an oligo(dT) column, for example). HOWEVER, it really depends what you're trying to amplify - if you're measuring ribosomal RNA for example, you DEFINITELY don't want to use an oligo(dT) as you'll not get any product! In that case, you'd either use random hexamers (to reverse transcribe everything) or a gene-specific primer (to only reverse transcribe your transcript of interset).
The main thing to remember is that RT-PCR is a VERY sensitive technique - theoretically, it can detect one molecule of RNA in a sample. Provided your RNA is intact and present in your sample, and you've used the right reverse transcription primer/PCR primers, you should be able to detect even very low concentration samples with total RNA. If you can't detect something you know should be there, it can often be a sign of a problem (e.g. RNA degradation).
I do agree with the two above comments that you go for all RNAs. Remember, when you use a commercial spin column RNA extraction kit the any piece of RNA even miRNA will be extracted. Good luck.
Agreed - on the basis of your experimental design purify total RNA and do your cDNA synthesis with an oligo(dT) primer - you'll get all RNA down but specifically reverse-transcribe mRNA. Most of my qPCR was done in S. pombe (not the same fungus, but a fungus nonetheless!) and this is how I always did it.