World records for drill-and-blast tunnelling from Norwegian contractors, bear witness to numerous weeks of more than 100m, and an exceptional 5.8 km in 54 weeks, also from one face. Earlier hard-rock world records using high-powered TBM in Norway, but most frequently and more recently, the records with Robbins TBM through non-abrasive limestones in the USA, provide numbers in meters per day, per week, and per month, which are of course, even more remarkable. Unfortunately there are contrary and undesirable TBM records, which are occasionally recurring events so not records, which see TBM stopped for months or even years in fault zones, or permanently buried in mountains. The many orders of magnitude range of performance suggest the need for better investigations, better choice of TBM, and better facilities for improving the ground ahead of TBM, when probe-drilling indicates that this is essential. Control of water, and improved stand-up behaviour in significant weakness zones and faults may demand drainage, which can be unending, and pre-injection. Fortunately there are increasing signs that this is recognized by TBM manufacturers: more guide-holes for drilling pre-injection umbrellas are seen through front-shields nowadays. A little acknowledged fact is that when all hours are included, TBM will generally decelerate as tunnel length and time increases. This is usually seen after improved performance during the learning curve. Deceleration is also a general trend during world-record setting performances. This means that utilization U is equal to the ratio of actual advance rate and penetration rate, AR/PR, only for specified time intervals, because U is time-dependent. This is rarely quantified by designers, and is therefore a source of risk, by default. Another important item for correct prognosis is the recognition that reduced penetration rate PR can sometimes occur when thrust is increased by the TBM operator, due to exceptionally resistant rock mass formations. Each of the above, and PR sensitivity to a wide range of cutter forces, UCS and abrasiveness, are provided in the empirical QTBM method. This method explains variable progress in jointed rock, which is sometimes fast, and also quantifies the likely delays in untreated, or pre-injected, fault zones.