If and only if you're an experienced project manager, and you've done similar projects before, with the same people and the same resources, then an educated guess will be the simplest but not necessarily the most analytic.
Sometimes a company will keep records of projects that have been completed. If you know what you're looking for, and the records are accurate, then this method of assigning times to activities may be better than just guessing.
Suppose you look to the past and there are several such projects and each of them have different times for similar activities in the project. In one case, Murphy must have been working overtime since the project came in way behind schedule and over-budget. Another one appears to have been done by a bunch of hot shots. It came in ahead of schedule and under budget. However several projects appear to have been done by ordinary mortals. They came in on time and on the money. To get a weighted estimate, use the following formula for each activity:
Time(opt) + 4*Time(likely) + Time(pess) Time(expected) = --------------------------------------- 6
where Time(opt) is the optimistic time, Time(pess) is the pessimistic time and "6" is just the sum of 1+4+1.
You'll need to brush up on computing the statistical variance for each task on the critical path and the statistical deviation for the project as a whole. Then, you can figure out Z-scores and figure out the probability that you'll complete a project in a certain number of days. You can then "set the odds" and take bets! The better thing to do is get a PM program that does all of this for you. You just plug in the numbers and see what happens. Of course you can fudge to your heart's content: no one will know where the numbers come from, but they'll look mighty impressive.