Debris disks are believed to be the remnants of planet formation; a disk of solid bodies called planetesimals that did not get incorporated into planets. They provide an ideal opportunity for studying the outcome of planet formation in their systems. The best studied disks exhibit cool emission peaking at ≥ 60 microns, lying in Edgeworth-Kuiper belt-like regions with an inner dust-free hole. However around half of the main sequence stars with excess emission seen in IRAS observations show an excess at 25 microns only. This thesis presents a study of mid-infrared debris disks through theory and observations to examine the following questions: are such disks around Sun-like stars simply debris disks of truncated planetary systems?; can this emission be explained by a collisionally-evolving disk analogous to the asteroid belt?; is the degree of variation in emission levels seen around otherwise similar A stars evidence of stochastic evolution? An analytical model of debris disk evolution assuming the disk evolves under a steadystate collisional cascade is presented and shows there is a maximum flux that can be expected from a disk of a given radius and age and that, for a given disk location, the excess emission arising from the disk will decrease linearly with time. Comparison of observations with the maximum predicted flux from the analytical model indicates some Sun-like stars are likely hosts of transient emission. Comparison with A star statistics shows that A star excesses can be explained by collisionally-evolving disks, and that the variation in emission between similar stars can be explained by varying initial conditions. However the model assumes the disk consists of a narrow ring (whereas the true dust distribution may be spatially extended) at a location predicted by blackbody fitting to the excess SED (which can lead to errors in the dust location of up to a factor of 3). Resolved imaging is needed to determine the true disk morphology and the implications of this on the transient or steady-state interpretation. A sample of 12 Sun-like stars with mid-infrared excess, and a complementary sample of 11 A-type stars, are observed with TIMMI2, VISIR on the VLT and MICHELLE and TReCS on Gemini. Six of the Sun-like sources are shown not to be debris disks, highlighting the need for high-resolution imaging to remove bogus disk sources. None of the Sun-like stars show resolved emission, however a new method of determining extension limits from unresolved imaging is presented and used to show that a single-temperature dust model for the η Corvi mid-infrared excess with transient dust at 1.7AU is more likely than a 2-temperature fit with dust belts at 1.3AU (transient) and 12AU (steady-state). The A star observations reveal a further bogus disk source. Unresolved images of HD71155 constrain the excess emission to be from 2 dust populations: a transient population at 0.6AU and a steady-state population at 61AU. The extension limits modelling is further used to highlight A star disks which may present fruitful subjects for future 8m imaging. One such source, HD181296, is observed with TReCS and shown to possess an edge-on disk at around 22 AU which fits with the steady-state interpretation. As the unresolved 8m observations are used to constrain the outer limits of the disk emission, MIDI observations of 2 Sun-like and 2 A-type sources are used to constrain the inner limits of the disk. The first results from these observations indicate that there are changes in the visibility function with wavelength that match the predicted changes for a completely resolved disk component. The combined limits for the hot Corvi emission suggest the source has a transient disk lying between 0.9AU and 3.0AU.