Long-term fire experiments in savannnas are rare, given the difficulties and demands of operation. Controlled fire experiments date from colonial times in West Africa, although the largest and best-known is located in the Kruger National Park, South Africa. The achievements of these experiments are assessed from examples in Africa, South America and Australia. A less well-known experiment in Zimbabwe was sited at the Marondera Grassland Research Station and ran from 1953 to 1991. Some of the preliminary results on the impact of fire on vegetation are analysed and compared with further vegetation surveys in 2007. Studies on tree growth in this miombo savanna woodland indicate that the plots burned at three- and four-year intervals recovered to greater mean heights than the unburned control plots. There was no significant variation between treatments, suggesting that the few trees that did survive in the frequently burned plots were large specimens. Brachystegia and Julbernadia dominated the plots throughout and after the experiment. Basal area and stocking density were highest in the four-yearly burned plots but there was a high variability throughout the experiment, suggesting that many trees may have attained heights and bark thicknesses sufficient to protect from fire damage. Fire also affected the composition of the herbaceous plant community, but not the number of species. By the end of the experiment some grass and sedge species had flourished while others revealed greater susceptibility to fire, and fire-tolerant species predominated in the most frequently burned areas. The experimental design appeared to cope well with the variability between plots and indicated the soundness of the initial design and its implementation.