Putting a number on work nobody has done yet, and saying honestly how confident you are in it.
An estimate is a probability distribution that gets reported as a single number, and most estimation trouble begins at that conversion. These books cover the counting, computing and judgement techniques, the outside view that corrects for optimism, and the flow data that lets you forecast without estimating at all. They suit anyone who has to commit to a date and would rather commit to a range with a stated confidence behind it.
McConnell separates the estimate from the target and the commitment, then gives you counting, computing and judgement techniques with published accuracy data attached to each. The cone of uncertainty and the chapter on presenting an estimate are the parts you will use in your next planning meeting.
Reference class forecasting, which means anchoring on how similar projects actually turned out rather than on the plan in front of you. Flyvbjerg's database of more than sixteen thousand projects is the evidence that the outside view beats the inside view almost every time.
Forecasting from cycle time percentiles and Monte Carlo simulation, so the answer comes from what your team has already done rather than from what it hopes to do. The route to a defensible date that never asks anyone to size a story.
Cohn separates estimating size from deriving duration, and works through the release, iteration and portfolio horizons in turn. The chapters on velocity and on communicating a plan under uncertainty have aged well.
DeMarco and Lister on why a single date hides the risk that produced it, and how a risk diagram turns that date back into a distribution. The uncertainty diagram changes what a sponsor asks you for next time.
Goldratt on the padding people add to every task and lose anyway, and on pooling it into a project buffer instead. The explanation of student syndrome is the clearest account of why task level estimates rarely protect a schedule.
Brooks on why effort and duration are not interchangeable, and on the scheduling arithmetic that follows once you accept it. The chapter on the tar pit is an honest description of how a small slip becomes a crisis.
The planning fallacy, anchoring and overconfidence, described by the researcher who documented them. Read it to understand why your team will repeat the same estimating mistake even after working through everything above.