This study assesses the performance of Botswana's traditional arable agriculture for the 1968-90 period, Growth rate and arable sub-sector production models are specified and estimated to determine how the sub-sector performed over time, and to capture the impact of the Accelerated Rainfed Arable Programme (ARAP). Growth rate model results indicate that cultivated area increased by about 2.2% per year during the 1968--90 period. However, crop output remained unchanged and yields declined by about 6. I% per year during the review period. Sub-sectoral model results reveal that cultivated area, output and yields rose by about 27%, 120% and 74% (respectively) due to the implementation of ARAP. Therefore, ARAP was effective in improving rural household food security and welfare. However, it is further argued that the program was unsustainable since it involved phenomenal government outlays and has led to an unprecedented input substitution from animal traction to tractor traction, which seems to be unjustified given the current economic fundamentals of the country's traditional arable farming. Moreover, the results reveal loss of productivity in the sub-sector over time. Therefore, the challenge facing policy makers is to devise new ways of reversing the current trend. (D