Honestly, this could be answered by searching pubmed (or even wikipedia) a bit. While miRNAs generally target the 3' UTR, they don't have to. Targeting of UTRs just allows more flexibility in targeting across tissues and developmental stages (you can have multiple UTRs without mucking with the protein sequence). Just so that there's a reference here, have a read through this open access paper, found by a trivial search of pubmed, that looks at how functional coding region target sites are.
The "classic" miRNA binding sites are located in the 3'UTR part of the mRNA.
Depending on the binding type they can destabilize the mRNA, cleave or induce translation suppression.
However, as dpryan79 already mentioned there are bona fide binding sites in the CDS region.
The latest version of the microT algorithm (DIANA-microT-CDS) can identify binding sites in 3'UTR and CDS regions as well.
You can find the tool at http://www.microrna.gr/microT-CDS/
More information about the last web server version (front-end) you can find in this open NAR paper: http://nar.oxfordjournals.org/content/41/W1/W169
For so many miRNAs, i would've downloaded the csv that is provided by microT-CDS. There you get miRNA, gene, chromosome, coordinates and interaction score.
Then you can filter the results based on your miRNAs/genes list and score of preference [0 is worse and 1 is best].
miRNAs can also bind on non-coding RNAs (there is lncBase in the same server for validated and predicted miRNA:lncRNA interactions).