The combined effect of many stars within the lensing galaxy is to produce a maze of caustics, elongated regions of high magnification with dimensions of microarcseconds which form an intricate pattern across which the source moves. Sources with angular sizes smaller than the characteristic scales of this pattern suffer time-dependent magnification as the pattern moves across them, and consequently the brightness of each lensed image varies as its line of sight crosses the caustic pattern. The details of the resulting effects on the image lightcurves were calculated in the years following the discovery of Q0957+561 (Chang & Refsdal 1979, Paczynski 1986, Kayser et al. 1986, Kayser & Refsdal 1989). It was first observed by Irwin et al. (1989) in the lens system Q2237+0305 (the "Einstein cross", Huchra et al. 1985) which is a four-image lensed system produced by a low-redshift spiral galaxy with a high central stellar density around the lensed images; the system is also useful because the time delays are small, much less than the timescale of variations due to microlensing.
The most basic information carried by the microlensing lightcurves is a combination of source size and mass of the microlensing objects (Schmidt & Wambsganss 1998, Wyithe et al. 2000, Yonehara 2001, Kochanek 2004). However, the fact that sources of different sizes respond differently to microlensing by the stars in the lens galaxy offers an opportunity to study sources in great detail (Wambsganss & Paczynski 1991), as well as a way to infer the presence of microlensing by differences in spectra between one image and another (e.g. Wisotzki et al. 2003).
The central region of quasars contain an accretion disk close to the central supermassive black hole, with temperatures of over 10000 K and producing hard UV and X-ray emission. Further from the nucleus are found broad-line regions, showing typical velocity widths of a few thousand km s-1; reverberation mapping studies of local broad-line AGN have yielded typical size scales of a few light-weeks for these areas. On larger scales still are likely to lie tori of material which reprocess the quasar radiation and re-emit photons in the infrared, together with narrow-line emission regions a few hundred parsecs from the centre.
Interesting results began to emerge about a decade ago as lensed quasars were monitored extensively at optical wavebands. It is expected that the source sizes are different in different optical colours, because the temperature of the accretion disk increases as its radius decreases, and this should show up as a chromatic microlensing signal (Wambsganss & Paczynski 1991) which was duly found in observational programmes (e.g. Wisotzki et al. 1993, Claeskens et al. 2001, Burud et al. 2002). This effect can be used to estimate the accretion disk size and structure (Poindexter, Morgan & Kochanek 2008, Morgan et al. 2008, Poindexter et al. 2010, Hutsemekers et al. 2010, Dai et al. 2010, Blackburne et al. 2011, Muñoz et al. 2011); the picture that emerges in some cases is of a scale-size of a few light days and a temperature-radius profile that is consistent with a standard Shakura-Sunyaev thin disk (Poindexter et al. 2008). But this is by no means a universal result, and in many cases the inferred size is bigger or the temperature profile is different. For example, Blackburne et al. (2011) analyse multiwavelength observations of a sample of lensed quasars and find that the microlensing properties of many of the objects imply accretion disk sizes of up to a factor of 10 larger than standard disks.
Comparison of the spectra of the broad emission lines in different images of quasar lens systems have shown that the BLR is also microlensed (Abajas et al. 2002, Richards et al. 2004, Wayth et al. 2005, Keeton et al. 2006, Abajas et al. 2007, Sluse et al. 2007, Hutsemekers et al. 2010, Sluse et al. 2012) as originally predicted 30 years ago (Nemiroff 1988, Schneider & Wambsganss 1990). Like the continuum microlensing studies, these are very important clues to the structure of the emitting object. Results from this work include the determination of the overall size scale of the BLR, ranging from < 9 light days in SDSS J0924+0219 (Keeton et al. 2006) to a few light-months in Q2237+0305 (Wayth et al. 2005), but lack of consensus on the structure of the BLR. This is obtained from the microlensing signature from different velocity components within each line; in some cases there is evidence for an ordered, biconical structure (Abajas et al. 2007) but larger surveys (Sluse et al. 2012) seem to show microlensing signals which are largely independent in red and blue wings, suggesting a non-spherical structure for the BLR.