Fisheries in African reservoirs are typically multi -species and in most cases the fish resource is harvested with a number of gears. These characteristics complicate their management and the development of management procedures. Typically, long time series of data on catch and effort and length- or age-based catch are not available for these fisheries. This precludes the use of data intensive methods such as multi-species virtual population analysis. The principal aim of this thesis was to develop a management procedure for African reservoir fisheries that takes into account the pertinent biological characteristics of the target species and accounts for the multi-species and multi-gear irIteractions in such fisheries. An opportunity availed itself to undertake this work on Lake ChicaIllba (19°08'S 33°08'E) a man-made hydroelectric dam in subtropical Mozambique (Manica province). The specific objectives of this study were: to obtain locality specific biological parameters for the target species in Lake ChicaIllba; to assess gear utilisation trends in the fishery through the determination of gear-selectivity, catch rate and effort for each of the principal gears used in the fishery; to assess the fishery using traditional per-recruit models and to test existing and new per-recruit models that account for the multi-species and multi-gear nature of the fishery and to determine the adequacy of each of these approaches in the determination of suitable target reference point (TRP) exploitation rates. The three principal specIes in Lake Chicamba are the introduced largemouth bass, Micropterus salmoides, and two cichlids the Mozambique tilapia, Oreochromis mossambicus and the redbreast tilapia, Tilapia rendalli. Sectioned otoliths were used for age and growth determination. Marginal zone analysis showed that annulus formation in all three species occurred during winter. The maximum-recorded age was 5 years for M. salmoides, 16 years for T. rendalli and 10 years for O. mossambicus. Growth of the three species was best described by the 3 parameter von Bertalanffy growth model as ℓa = 465.51 (1 - e⁻ₑ·ₑ⁷⁵⁽a⁺⁰·⁰⁰⁹⁾ mm FL for M salmoides; ℓa = 238.74 (1 - e⁻⁰⁶³⁶⁽a⁺⁰·⁹⁰⁵⁾) mm TL for T. rendalli; and ℓa = 266.06 (1 - e⁻⁰⁷⁹⁰⁴⁽a⁺⁰·²⁶⁹⁾) mm TL for 0. mossambicus. Female T. rendalli attained 50%-maturity at 2.89 years, while O. mossambicus matured at 2.83 years and M. salmoides at 0.9 years. Both cichlid species spawned throughout summer while M. salmoides had a very short spawning season from August to September. The total annual mortality rate (Z) for M. salmoides in Lake Chicamba was 1.27 yr⁻¹, the mean empirical estimate of natural mortality (M) was 0.73 yr⁻¹, and fishing mortality (F) was calculated at 0.54 yr⁻¹. For T. rendalli Z = 0.31 yr⁻¹, M = 0.20 yr⁻¹, F = 0.11 yr⁻¹ and for 0. mossambicus Z= 0.62 yr⁻¹, M= 0.38 yr⁻¹, F= 0.24 yr⁻¹. The three species exhibited reproductive traits, which implied a high reliance of recruitment on spawner stock (nest guarding in T. rendalli and M. salmoides and mouthbrooding in O. mossambicus). For this reason it was decided that the cichlid fisheries should be managed using TRPs which maintained the spawner biomass-per-recruit at 50% (FSB50) of pristine levels. However, based on good evidence it was hypothesised that the high rate of fishing mortality helped to maintain the fast growth rate of M. salmoides. It was, therefore, decided to manage this species at a TRP of F SB40. The three most important fishing sectors were the gill-net, seine-net and hook-and-line fisheries. The total catch for 1996 was 223 t. The gill nets selected all three species at a size/age approximating 50%-maturity while the seine-net and hook-and-line fisheries selected mainly juvenile fishes. There was strong evidence to suggest that seine net fishing also disrupted spawning. It was shown that the 'traditional' single-species per-recruit models were unsuitable to assess multi-species and multi-gear reservoir fisheries. Since existing multi-species/multifishery yield-per-recruit models were not capable of defining FsB(x) TRPs, a new multispecies/ multi-fishery spawner-biomass-per-recruit approach was developed. This approach allowed for the simulation of the response of spawner biomass-per-recruit to changes in effort in the three fishery sectors, simultaneously. The models showed that the spawner biomass-per-recruit, at current effort levels, was higher than the suggested TRP for the three species. However, it was shown that an increase of 10% in current total effort would reduce spawner biomass-per-recruit to below the recommended TRP levels. With the closure of the seine-net fishery, gill-net effort could be increased to 338 fishers (340 for management purposes) and effort in the hook-and-line fishery could be increased by 30% before the TRP was reached. To maintain the fish stocks above TRP levels, effort control was considered to be the most effective management method. The main recommendations for Lake Chicamba were to close the seine-net fishery, to limit the gill-net fishery to 340 fishers (using 137-m long x 3-m deep gill nets) and to maintain the open access nature of the hook-and-line fishery. The multi-species/multi-fishery per-recruit approach allows for the meaningful simulation of various scenarios and provides relatively robust management options. In the absence of long time series of effort and age- or length-based catch data, this approach was considered as the most suitable assessment method for multi-species/multi-gear African reservoir fisheries.