This review uses a candidate gene approach to identify possible pharmacogenetic modulators of opioid therapy, and discusses these modulators together with demonstrated genetic causes for the variability in clinical effects of opioids. Genetically caused inactivity of cytochrome P450 (CYP) 2D6 renders codeine ineffective (lack of morphine formation), slightly decreases the efficacy of tramadol (lack of formation of the active O-desmethyl-tramadol) and slightly decreases the clearance of methadone. MDR1 mutations often demonstrate pharmacogenetic consequences, and since opioids are among the P-glycoprotein substrates, opioid pharmacology may be affected by MDR1 mutations. The single nucleotide polymorphism A118G of the mu opioid receptor gene has been associated with decreased potency of morphine and morphine-6-glucuronide, and with decreased analgesic effects and higher alfentanil dose demands in carriers of the mutated G118 allele. Genetic causes may also trigger or modify drug interactions, which in turn can alter the clinical response to opioid therapy. For example, by inhibiting CYP2D6, paroxetine increases the steady-state plasma concentrations of (R)-methadone in extensive but not in poor metabolisers of debrisoquine/sparteine. So far, the clinical consequences of the pharmacogenetics of opioids are limited to codeine, which should not be administered to poor metabolisers of debrisoquine/sparteine. Genetically precipitated drug interactions might render a standard opioid dose toxic and should, therefore, be taken into consideration. Mutations affecting opioid receptors and pain perception/processing are of interest for the study of opioid actions, but with modern practice of on-demand administration of opioids their utility may be limited to explaining why some patients need higher opioid doses; however, the adverse effects profile may be modified by these mutations. Nonetheless, at a limited level, pharmacogenetics can be expected to facilitate individualised opioid therapy.