Multiscale base-scale entropy （MBE） is used to analyze heart beat interval series of both congestive heart failure （CHF） sufferers and the healthy people. The results indicate that, in contrast to the higher base-scale entropy（BE） value in short time scale, the data of CHF sufferers have actually lower BE value compared with those of the healthy people in long time scale. This supports the view that CHF patients’ heart beats are more random in short time scale due to impaired cardiac dynamic system, but are not versatile enough in long time scale to adapt to external impacts. More importantly, we found that δ, a parameter in MBE analysis, can be used to distinguish effectively the CHF patients from the healthy, which is proved in analysis of a database containing 72 healthy people and 44 CHF sufferers.