To analyse the relationships among fish species performance and management procedures, a database was built up with data from 31 fish farms during the period 1976–1987 (1673 observations) and analysed through multivariate statistics (factor analysis). The data include nurseries, grow-out and operational ponds with mono- and polycultures of common carp, Cyprinus carpio, tilapia hybrid, Oreochromis niloticus × O. aureus, silver carp, Hypophthalmichthys molitrix, and mullet, Mugil cephalus. The main conclusions include the following. (1) The highest total yields and best tilapia performances were obtained in polyculture ponds where tilapia was the main species. (2) The best carp performances occurred in grow-out polyculture ponds where carp was the main species. Carp performance was improved in polycultures with mullet and silver carp, irrespective of whether tilapia were present or not. (3) Carp and tilapia yields increased as the nutritional inputs (feed pellets, sorghum pellets, manure), pond size and culture duration increased. The effect of the nutritional input was not linear, but logarithmic. (4) Growth rate of common carp was more affected by total density and stocking size than that of tilapia. Better carp and tilapia growth occurred in grow-out ponds when stocked at large sizes and cultured during short periods, mainly when both species were present. (5) Carp growth varied with the geographical region and size of fish pond, being better in smaller than in larger ponds due to reduced access to natural benthic food in deep ponds.