The aim was to evaluate the relative importance of an established battery of five function tests used in the assessment of cardiovascular autonomic neuropathy (CAN) in diabetes employing spectral analysis of heart rate variability (HRV) as the reference test. In a population-based sample of 178 diabetic persons and 194 non-diabetic controls, five function tests (E/I ratio, Valsalva, 30:15 ratio, Orthostatic-BP, and sustained Handgrip) and power spectral analysis of a 24-h Holter recording were carried out. The high-frequency power during nighttime (LnHF(NIGHT)) was taken to express the parasympathetic function and the daytime low-frequency power (LnLF(DAY)) to express the sympathetic function. The readings were log transformation when appropriate, age-corrected, and standardized to units of standard deviation. Combinations of the three mainly parasympathetic tests and the two mainly sympathetic tests were computed by averaging the standardized readings. A high value of the mean sum of squares in LnHF(NIGHT) or LnLF(DAY) - explained in regression analysis - was assumed to represent the better test or combination of tests. The three parasympathetic function tests each correlated significantly to LnHF(NIGHT). The E/I ratio had a correlation to LnHF(NIGHT) similar to the combination of the three parasympathetic tests and the combination of all five function tests, whereas Valsalva and 30:15 ratio had a significantly poorer association. Sustained Handgrip-correlated significantly poorer to LnLF(DAY) compared to Orthostatic BP and the combination of the two sympathetic tests explained a significantly smaller part of the variation in LnLF(DAY) and LnHF(NIGHT) than did the combination of all five tests. This study indicates that: (1) no information is gained by adding the sympathetic function tests to the parasympathetic tests, (2) the most informative test is the E/I ratio, (3) and knowledge about the degree of CAN as defined by reduced HRV is not significantly increased when the four other function tests assessed are added to the E/I ratio.