This paper briefly surveys several conceptions of naturalness in phonology, touching primarily on typological frequency and the notion of 'phonetic motivation'. It is argued that typological frequency is not a reliable indicator of what is 'phonetically motivated' as relative frequency patterns are the outcomes of more complex interactions including non-phonetic factors. Phonetic motivations are diverse and include random variations, not only deterministic results, as is often desired. Models that view phonological patterns as emerging from complex interactions of a variety of natural factors are the most satisfying.