For various experiments (e.g. ALICE, CMS etc) at the CERN-LHC, people talk about tracking efficiency. What exactly is this tracking efficiency? How different is it from reconstruction efficiency?
Charged particles are detected in 'tracking detectors' or 'trackers'. These may be gaseous drift chambers, scintillating fibres, or silicon detectors but they share the property of measuring the position of the particle in a series of 'hits', detecting and amplifying the small amounts of energy it loses to exciting/ionising atoms in the material it passes through. They do not measure the energy of the particle but (usually) they let you determine the momentum from the curvature of the track in the strong magnetic field which is applied.
Ideally the software takes these hits and reconstructs the helical track trajectory from them ('join the dots...'). In practice this is limited by the intrinsic efficiency of the detector, possible non-working parts of the system, the amount of background noise and how good a job the software does. These combine to give the tracking efficiency - the fraction of tracks that will be found and reconstructed. It depends on the track's momentum, its direction, running conditions and anything else you can think of. It is generally pretty good - well above 90% - unless the track has a very low energy or is heading down the beam pipe or into some other region of low acceptance.
Tracking efficiency is part of reconstruction efficiency. If you are looking for a particular reaction (say $H^0 \to \mu^+ \mu^+ \mu^- \mu^-$) then to detect an event you have to measure all 4 tracks and identify them as muons, so the reconstruction efficiency would involve the 4th power of the tracking efficiency and the 4th power of the efficiency for muon detection (which is another story), and whatever cuts the physicist imposes in the analysis to reduce backgrounds. Reconstruction efficiencies are specific to each reaction and each analysis: the skill of the analyst is to devise a selection method in which the reconstruction efficiency is high and the background low, and both these quantities are well determined and not subject to uncertainties.