1. Cyclosporine (CsA) concentrations in plasma and whole blood were monitored prospectively in 66 consecutive kidney transplant recipients for 6 months after transplantation or until graft loss. Immunosuppression was based on treatment with CsA and prednisolone in 27 patients and CsA, azathioprine and prednisolone in 39 patients. 2. Whole blood and plasma samples (separated at 37 degrees C) were collected 10‐12 h after CsA dosage twice weekly over the first 3 months and thereafter once weekly. CsA concentrations were measured by high pressure liquid chromatography (h.p.l.c.) in plasma, by specific and non‐specific monoclonal radioimmunoassays (r.i.a.) in whole blood, and by polyclonal r.i.a. and polyclonal fluorescence polarization immunoassay (f.p.i.a.) in whole blood and plasma. 3. There were no differences between the treatment schedules regarding graft or patient survival, occurrence of acute rejection, nephrotoxicity or infection. 4. CsA concentrations were significantly lower at the time of acute rejection than one week earlier based on all of the analytical methods used except f.p.i.a. 5. The lowest CsA concentration, recorded during the first month after transplantation, was significantly lower in patients with than in patients without experience of acute rejection episodes when the CsA concentrations were measured by polyclonal r.i.a. in whole blood and plasma and by specific and non‐specific monoclonal r.i.a. in whole blood, but not by h.p.l.c. in plasma or polyclonal f.p.i.a. in whole blood or plasma. 6. The highest CsA concentration recorded during the second post‐transplantation month, was higher in patients with acute nephrotoxicity than in those without nephrotoxicity when CsA was measured by specific monoclonal r.i.a. in whole blood (471 +/− 409 ng ml‐1 vs 327 +/− 150 ng ml‐1, P less than 0.05), but not by the other methods. 7. The mean plasma h.p.l.c. concentration of CsA measured by h.p.l.c. during the first month after transplantation was significantly higher in patients who suffered from systemic infection than in patients who did not (116 +/− 70 ng ml‐1 vs 82 +/− 52 ng ml‐1; P less than 0.05). 8. Thus, significant relationships between CsA concentrations and clinical events were apparent using assay methods specific for CsA as well as using polyclonal r.i.a., but not using polyclonal f.p.i.a. When h.p.l.c. was used, however, plasma drug concentrations were often below the limit of determination. Our results suggest that specific analysis of CsA in whole blood allows the best distinction between patients who respond favourably and less favourably to treatment with CsA.