Individuals vary in their ability to perceive, as conscious sensations, signals like the beating of the heart. Tests of such interoceptive ability are, however, constrained in nature and reliability. Performance of the heartbeat tracking task, a widely used test of cardiac interoception, often corresponds well with individual differences in emotion and cognition, yet is susceptible to reporting bias and influenced by higher-order knowledge, e.g., of expected heart rate. The present study introduces a new way of assessing cardiac interoceptive ability, focusing on sensitivity to short-term, naturalistic changes in frequency of heartbeats. At rest, such heart rate variability typically reflects the dominant influence of respiration on vagus parasympathetic control of the sinoatrial pacemaker. We observed an overall tendency of healthy participants to report feeling fewer heartbeats during increases in heart rate, which we speculate reflects a reduction in heartbeat strength and salience during inspiratory periods when heart rate typically increases to maintain a stable cardiac output. Within-participant performance was more variable on this measure of cardiac interoceptive sensitivity relative to the “classic” heartbeat tracking task. Our findings indicate that cardiac interoceptive ability, rather than reflecting the veridical monitoring of subtle variations in physiology, appears to involve more interpolation wherein interoceptive decisions are informed by dynamic working estimates derived from the integration of afferent signaling and higher-order predictions.