The population of B. mattozi in the Mtsheleli Dam, Zimbabwe, was estimated to be about 55 000 fish in July 1974 but fell to about 370 by June 1975. The total biomass followed a similar trend, reaching 2 000kg (43kg ha–1) in July and falling to 130kg (2.8kg ha–1) after one year. The mortality rate was very high (Z = 5.00) and considerably higher than in a number of other fishes, many of them considerably smaller than B. mattozi. The reasons for this decline are discussed and they include the effects of humans who removed about 17.5% of the total population, cannibalism, otters and birds. A remarkable feature of the population was the lack of small fish (<12cm FL), which was attributed to a failure of spawning during a serious drought in 1972–1973. The numbers of small fish began to increase from June 1975 and these were most likely the cohort spawned during the good rains of 1973–1974. Fluvial fishes in semi-arid parts of southern Africa are likely to experience wide variations in numbers determined by the success or failure of spawning in wet or dry years.