INTRODUCTION: The correlation dimension (CD) method is related to chaos theory and is used to quantify heart rate variability (HRV). The CD is a measure for the amount of correlations present in the signal. METHODS: The algorithm used to calculate the CD is based on the method of Grassberger and Proccacia. The method was first validated on signals with known CD and then applied to HRV-signals of heart transplants and an age-matched control group of healthy subjects. The CD of the corresponding surrogate time series was calculated to investigate non-linear correlations in the HRV-signal. Circadian variations of the CD were studied in 20 healthy subjects, including men and women. RESULTS: The value of the CD for healthy subjects ranged from 2.12 to 5.53 with a mean value of 4.32. For heart transplants, only a few time series showed a finite value of the CD that varied between 2.10 and 5.60. Also, a significant difference was found between the CD of the original and the surrogate time series in healthy subjects. The CD of women is higher than the CD of men, and this difference was more pronounced during the night than during the day. CONCLUSIONS: This limited study shows that the CD alone cannot be used to make a distinction between HRV-signals of healthy subjects and of heart transplants. However, there is evidence that there are non-linear correlations present in the HRV-signal and that there are significant gender and circadian differences in the CD.