Microcells deployed inside buildings typically do not have a dedicated spectrum allocated to them. Spectrum scavenging is a scheme that has been proposed in Grandhi et al. (1995) for identifying frequencies from the macrocell spectrum for use in microcells deployed inside buildings. To start up service in a microcell, a set of frequencies for use in the microcell is identified by the spectrum scavenging procedure of that microcell running in the start-up phase. After a set of usable frequencies are identified the spectrum scavenging process is relegated to the background. A dynamic channel assignment scheme is used to assign channels (TDMA time slots or AMPS frequencies) to users in the microcells. The spectrum scavenging process that is running in the background continually grooms the usable set of frequencies. We present the performance of the spectrum scavenging scheme by simulating indoor microcells embedded in a macrocell environment. The convergence rate, stability and adaptability of the spectrum scavenging algorithm are studied in this paper.