Abstract This study was designed to test a proposed dose modification for intravenous milrinone in congestive heart failure patients (CHF, NYHA I-II) with either moderate or severe renal impairment. All the patients were administered an intravenous loading dose of drug at 50 μg kg-1 over 10 min. This was followed by an 18 h maintenance infusion of milrinone at 0·45 or 0·35 μg kg−1 min−1 for the moderate (chromium-EDTA clearance of 31–75 mL min−1, n = 10) and severe renally impaired subjects (chromium-EDTA of clearance 10–30 mL min−1, n = 11), respectively. Plasma and urine samples were collected for up to 34 h and analysed for parent drug by validated HPLC methods. The mean (± s.d.) steady-state plasma concentrations of milrinone were within the therapeutic range (100–300 ng mL−1) for both groups, with values of 239 ± 71 ng mL−1 and 269 ± 32 ng mL−1 for the moderate and severe patients, respectively. No statistical differences were observed between the steady-state values for the two groups. With the exception of two patients per group, individual steady-state levels were also within the therapeutic range. Those outside the nominal range showed steady-state levels, ranging between 308 and 353 ng mL−1, that were not associated with any serious adverse events. As predicted for this highly renally cleared drug, there were differences (P < 0·001) in the total plasma clearance (CLP), renal clearance (CLr), and plasma terminal half-life (t1/2) of drug, with values in the severe group being 44% lower, 75% lower, and about 134% longer respectively, when compared with the moderate group. High (correlation coefficient > 0·8) and significant correlations (P < 0·001) were observed between CLP and CLr and the degree of renal impairment (chromium-EDTA clearance). The apparent volume of distribution was approximately 40% higher (P < 0·01) in the severe group compared with that for the moderate group (moderates were 0·443 ± 0·155 L kg−1). This volume difference suggests a decrease in the plasma protein-binding of milrinone because of the renal disease. The fraction of drug excreted in the urine was 0·705 ± 0·100 for the moderate group and 0·320 ± 0·089 for the severe group (P < 0·001). These results may suggest an increase in non-renal clearance of the compound, representing a partial compensation mechanism for the reduced renal function. In conclusion, this study has confirmed that the current dose reductions recommended for the use of intravenous milrinone in CHF patients with impaired renal function will yield plasma concentrations of the drug within the therapeutic range.