This paper examines biased technical change in South African agriculture using a system of share equations with unobserved components. Developing on the work of Lambert and Shonkwiler (1995), this paper generalises previous work by introducing independent unobserved components into each model using a regression-based approach. We find evidence of stochastic technical change, which is itself biased between the four factors of production: machinery, land, labour and fertiliser, and which closely reflects distinct phases of South African agricultural policy and development.