As a flexible maintenance strategy, Condition Based Maintenance (CBM) has been accepted by industry due to its efficiency and robustness in many engineering practices. Successful implementation of CBM relies on observation of actual health condition of machinery. Therefore, it is crucial to perform condition monitoring in CBM. This paper focuses on quantifying health condition of machinery. Empirical Mode Decomposition (EMD) is employed to decompose signal and extract dominant signatures, which could reflect health condition variation of machinery. Then, a novel index called Health Index (HI) is proposed to describe condition development trends. In order to detect occurrence of early faults, a dynamic threshold is also proposed. In case occurrence of early fault, HI should be higher than its corresponding threshold. This novel condition monitoring method is more appropriate for on-line health monitoring and detection of incipient fault. Two sets of data collected from gearboxes are used to validate the proposed method. The analysis results show that the proposed method is effective in condition monitoring, especially the detection of early faults.