The effects of liver disease on pharmacokinetics and pharmacodynamics are highly variable, and difficult to predict as the mechanisms of these effects are not well understood. Since the majority of the published literature is concerned with cirrhotic liver disease, this review also focuses mainly on this area. Four different theories have been proposed to account for the effects of chronic liver disease with cirrhosis on hepatic drug elimination: the sick cell theory; the intact hepatocyte theory; the impaired drug uptake theory; and the oxygen limitation theory. While some data in support of each of the first 2 theories have been published recently, a large amount of clinical data would appear to refute both of these theories. These clinical data are substantially consistent with the latter 2 theories, which regard the decreased permeability of the capillarised sinusoid as the critical feature in cirrhosis. Further work is required to determine the applicability of each of these theories. In cirrhosis, drug glucuronidation is spared relative to oxidative drug metabolism; however, in advanced cirrhosis this pathway may also be impaired substantially. There is evidence that in cirrhosis other conjugation pathways may also be impaired to variable degrees. Growing evidence suggests that biliary drug excretion is impaired in cirrhosis. Recent studies with several racemic drugs indicate that the disease can have different effects on the hepatic elimination of individual enantiomers, which may lead to a change in the concentration-response relationships of racemic drugs in cirrhosis. A major finding which has emerged in recent years is that, even with moderate degrees of hepatic impairment, there is a decrease in clearance of drugs or active metabolites normally cleared by the kidney. The effect on renal clearance of unbound drug may be masked if there is a concomitant decrease in plasma protein binding of the drug. Neither serum creatinine levels nor creatinine clearance are useful markers of the renal dysfunction associated with cirrhosis. Both may greatly overestimate renal function in patients with cirrhosis due to increased fractional renal tubular secretion of creatinine. Altered receptor sensitivity has been observed with some drugs in cirrhosis, while for other drugs there is no change in pharmacodynamics. Precise determination of drug dosage in cirrhosis requires information on changes in pharmacodynamics and plasma protein binding in addition to changes in drug elimination. Pharmacokinetic investigations in a variety of chronic liver diseases without cirrhosis (e.g. carcinoma, schistosomiasis and viral hepatitis) suggest that in the absence of cirrhosis, impairment of drug elimination is not sufficient to warrant reduction of drug dosage. However, if cirrhosis is present, 'safe' drug use requires an awareness of the possibility of multiple interactions between changes in hepatic and renal disposition and pharmacodynamics. In chronic liver disease with cirrhosis, dosage reduction is the general rule regardless of the route of elimination of drug or metabolite.