We've found ribosomal RNA to be less of a problem with sequencing that depends on polyA, which suggests the issue might be in the library preparation, rather than the selection.
Many polyA RNA library preparation methods involve amplification, rather than selection, which means that existing transcripts that are present in very high abundance (such as rRNA) will still show up in the mixture.
But polyA-tailed RNA isn't the only RNA present in a sample. For a comprehensive look at RNA, the whole-RNA sample preparation methods can't use any form of selection to identify target molecules. In this case, it's typically recommended to use a form of ribosomal RNA removal, where a probe to ribosomal RNA is attached to magnetic beads (or a column, or similar), and used to fish out out ribosomal RNA, leaving the remainder behind. Unfortunately those probes don't work perfectly, either due to the ribosomal RNA being damaged at the probe site, or the probe not matching the ribosomal RNA molecule, or the RNA just not getting into the right physical state to bind with the probe. All these things mean incomplete removal of ribosomal RNA, ending up with a situation where the highly-abundant ribosomal RNA still pokes through in the reads.
