The purpose of this study was to identify baseline procedures that are equally sensitive to negative and positive changes in heart rate variability, specifically in the context of moment-to-moment comparisons. Participants were engaged in a dual-task paradigm consisting of either a high- or low-workload primary task paired with a secondary task. Shifts between primary tasks occurred at set intervals. Average arousal was calculated for each primary task in relation to six different baseline procedures. A 2 � 6 repeated-measures ANOVA revealed a significant main effect for the baseline type, F(1, 58) = 3.90, p = .04, η² = .08. Planned comparisons revealed that certain procedures were biased to detect positive change and others negative change. The results of this study provide evidence that when developing physiologically based measures for detecting behavioral and/or cognitive state changes, the method used to calibrate these measures requires special consideration.