We propose a likelihood approach for events projection based on blinded or partially blinded survival data, using mixture of the underlying distributions (such as exponential or Weibull). The maximum likelihood estimates of the hazard rates and hazard ratio are then used to estimate the number of events and the statistical power by a future calendar time (e.g., at the planned termination of the study or at possible interim analysis times), taking account of both time to event of interest and time to censoring process, and conditioning on the known status (having had an event of interest, loss to follow-up, competing risks, administrative cutoff, etc.) of all the subjects at a given time. Plots of predicted number of events and power with associated 95% confidence intervals (derived using the delta method) vs calendar time are generated to facilitate the decision making. Real data examples from Phase 3 and Phase 4 diabetes clinical trials with cardiovascular risk assessment are used to illustrate the method.