This thesis describes the application of microrheology to characterize the mechanical properties of three soft matter systems: an entangled biopolymer solution, a suspension of actively swimming bacteria, and a gel-forming carbon nanotube network. We demonstrate using these distinct model systems that it is possible to employ microrheology to extract both local and bulk information using a combination of one- and two- point measurements and theoretical modeling. In the first set of experiments, we use microrheology to probe the rheological properties of semi-dilute polymer solutions of $\lambda$-DNA. In these solutions, the depletion interaction leads to a layer of reduced DNA density near the particle's surface. We demonstrate a method for deducing the local microstructure of these layers along with the bulk rheology of the polymer solution. This work was one of the first to systematically demonstrate that tracer-based microrheological methods could be used to deduce both local and bulk rheology in a well-characterized model soft matter system. In the second set of experiments, we use microrheology to probe the dynamics of a model active soft matter system: a suspension of swimming bacteria. By comparing measurements of the fluctuations of passive tracer particles with the response of a driven, optically trapped tracer in the bacterial bath, we demonstrate a breakdown of the fluctuation-dissipation theorem in bacterial baths. These measurements enable us to extract the power spectrum of the active stress fluctuations. We develop a theoretical model incorporating coupled stress, orientation, and concentration fluctuations of the bacteria to explain the observed scaling of the power spectrum. In the final set of experiments, we report measurements of gelling rigid rod networks, comprised of a semidilute dispersion of surfactant stabilized carbon nanotubes. Microrheology is employed to follow the rheological evolution of the suspension from a semidilute solution of unbonded tubes to a bonded gel network. A theoretical model based on the crossing probability of rods confined to finite volumes is developed to account for network elasticity. Model predictions compare well with computer simulations and experiments as a function of nanotube volume fraction and cure time.