The objective of this study is to evaluate the effect of exercise on cardiac autonomic function as measured by short-term heart rate variability (HRV) in females suffering from rheumatoid arthritis (RA). Females with confirmed RA were randomly assigned to an exercise group (RAE) and a sedentary group (RAC). RAE was required to train under supervision two to three times per week, for 3 months. Three techniques (time domain, frequency domain and Poincaré plot analyses) were used to measure HRV at baseline and study completion. At baseline, RAC (n = 18) had a significantly higher variability compared to RAE (n = 19) for most HRV indicators. At study completion, the variables showing significant changes (p = 0.01 to 0.05) favoured RAE in all instances. Wilcoxon signed rank tests were performed to assess changes within groups from start to end. RAE showed significant improvement for most of the standing variables, including measurements of combined autonomic influence, e.g. SDRR (p = 0.002) and variables indicating only vagal influence, e.g. pNN50 (p = 0.014). RAC mostly deteriorated with emphasis on variables measuring vagal influence (RMSSD, pNN50, SD1 and HF (ms(2)). Study results indicated that 12 weeks of exercise intervention had a positive effect on cardiac autonomic function as measured by short-term HRV, in females with RA. Several of the standing variables indicated improved vagal influence on the heart rate. Exercise can thus potentially be used as an instrument to improve cardiac health in a patient group known for increased cardiac morbidity.