The correlation between the daily dose of digoxin, its plasma level and clinical characteristics of 213 patients receiving beta-acetyldigoxin treatment has been evaluated. After logarithmic transformation of lognormally distributed variables multiple linear regression analysis was performed. Eight predictor variables were chosen: sex, age, height, weight, glycoside dose, creatinine and potassium level in serum and dose of spironolactone. The resulting correlation coefficient was 0.46, i.e. only 100 x r2 = 21.4% of the variance of the steady state digoxin plasma level could be interpreted with the aid of these variables. For only 4 of the 8 variables (age, glycoside dose, serum level of creatinine, and spironolactone dose) were partial coefficients of regression and of correlation significantly different from zero. Almost 80% of the variance could not be accounted for. This finding is in accordance with conclusions in the literature.