Time estimation of a given target is affected by the ensemble statistics of the set of intervals that it belongs to. Here we investigated this ensemble context effect using a bisection task in which target intervals were embedded in two types of sample distributions giving rise to low and high ensemble means and measured common EEG signatures of timing, contingent negative variation (CNV), P2 and late negative component of timing (LPCt). The CNV and LPCt components are interpreted to reflect anticipation and decision making mechanism, respectively, whereas P2 component is thought to index temporal accumulation. Therefore, we expected the contextual modulations to have an effect on the activities of both CNV and LPCt, but not on the P2. As predicted, we observed that intermediate durations were overestimated in short compared to long contexts and this bias was related to the CNV and LPCt activities, where the ERP negativities were higher and the offset positivities were lower. Moreover, the LPCt results indicated a linear decrease in amplitude by increasing target durations, suggesting an active decision making process at the stimulus offset for short target intervals. On the other hand, P2 amplitude linearly increased by target durations, while showing no context modulations. In general, the current study results suggest that temporal contexts induce different anticipation and decision making demands for the same stimulus durations and it is possible to study the underlying cognitive mechanism of temporal context by examining the CNV and LPCt amplitudes.