In many experimental set-ups, an ECG tracing is not available and the heart rate is evaluated from the inter-beat intervals (IBI) derived from the arterial blood pressure (ABP), often measured noninvasively at the finger site. However, whether any IBI series derived from ABP can be used instead of the R-R intervals (RRI) of the ECG to quantify the multiscale entropy (MSE) of heart rate variability is still unknown. Therefore we aim to quantify the discrepancies between MSE estimated using RRI or an ABP-derived IBI. For this purpose, we considered ECG and noninvasive finger ABP recorded for 15 minutes in 40 supine healthy participants. From the ECG we derived the RRI series; from the ABP we derived the series of systolic intervals (SSI), diastolic intervals (DDI), and intervals between maxima of ABP first- (dPI) and second- (dPI2) derivative. For each series, we estimated the MSE with embedding dimension m=2 deriving MSE at scale n=1 (i.e., SampEn) and at high-frequency (MSE-HF) and low-frequency (MSE-LF) scales, as described in [1]. We compared the estimates using RRI with the estimates using the ABP-derived IBI series by the Wilcoxon signed-rank test. Sampen was largely overestimated by the ABP-derived IBI series, the best performance being provided by dPI (percent error +11%), the worst by DDI (+16%). By contrast, MSE-HF was significantly but only slightly underestimated by the ABP-derived IBI series, the closest estimate being provided by dPI and SSI (-1%), the worst by d2PI (-2%); MSE-LF was also significantly underestimated, with the best performance by dPI and SSI (-2%), the worst by DDI (-3%). Therefore ABP-derived IBI series reflect entropy components not present in the RRI series particularly relevant in the SampEn estimate. Moreover, the ABP-derived IBI series that provides MSE estimates closest to RRI appears to be dPI. [1] Entropy, 2019 doi:10.3390/e21060550