Based on a quantitative understanding of the environmental factors effecting armoured bush cricket, Acanthoplus discoidalis population dynamics, a hypothesis was formulated to explain the occurrence of outbreaks in some years and locations. The principles, expressed using a rule-based or qualitative model, were that nymph and adult survival and fecundity were reduced in years with uneven rainfall and that egg survival was reduced in years with a wet late-season as this is associated with increased egg predation and/or water-logging. The implication was that large egg banks resulted either when a large number of adults were present or when neither low fecundity nor high egg mortality were constraints. Such large egg banks were however predicted only to lead to outbreaks when there was an adequate amount of food for nymph and adult survival in the following season. Model predictions were compared with observed outbreaks of A. discoidalis between 1988 to 2002 for the three climatic zones of the east, central and western parts of southern Botswana. There was significant agreement between model predictions and observed outbreaks in two of the three zones (95% confidence interval of the kappa coefficient of agreement > 0). Taking the data for all three zones together and compared to the average outbreak frequency, an outbreak was three times more likely to occur when the model predicted an outbreak and six times less likely to occur when it predicted no outbreak. Keywords: Pest forecasting, simulation, Rule-based model, Semi-arid.