Medusahead is one of the most problematic rangeland weeds in the western United States. In previous studies, prescribed burning has been used successfully to control medusahead in some situations, but burning has failed in other circumstances. In this study, trials were conducted using the same protocol at four locations in central to northern California to evaluate plant community response to two consecutive years of summer burning and to determine the conditions resulting in successful medusahead control. During 2002 through 2003 large-scale experiments were established at two low-elevation, warm-winter sites (Fresno and Yolo counties) and two higher elevation, cool-winter sites (Siskiyou and Modoc counties). Plant species cover was estimated using point-intercept transects, and biomass samples were taken in each plot. After 2 yr of burning, medusahead cover was reduced by 99, 96, and 93% for Fresno, Yolo, and Siskiyou counties, respectively, compared to unburned control plots. Other annual grasses were also reduced, but less severely, and broadleaf species increased at all three sites. In contrast, 2 yr of burning resulted in a 55% increase in medusahead at the coolest winter site in Modoc County. In the second season after the final burn, medusahead cover remained low in burned plots at Fresno and Yolo counties (1 and 12% of cover in unburned controls, respectively), but at the Siskiyou site medusahead recovered to 45% relative to untreated controls. The success of prescribed burning was correlated with biomass of annual grasses, excluding medusahead, preceding a burn treatment. It is hypothesized that greater production of combustible forage resulted in increased fire intensity and greater seed mortality in exposed inflorescences. These results demonstrate that burning can be an effective control strategy for medusahead in low elevation, warm-winter areas characterized by high annual grass biomass production, but may not be successful in semiarid cool winter areas.