Sudan's irrigated subsector is the largest in sub-Saharan Africa. Farming is practised under a scheme-mandated rotation with highly centralized decision making. Under this system, labour is the major input for which the tenant has allocation flexibility both during the season and across the three crops grown, sorghum, ootton, and groundnuts. This paper analyzes the risk attributes of the production technology and measures farmer's attitudes towards risk in the irrigation schemes of Sudan. Stochastic production functions are specified where risk increasing and risk reducing input effects are allowed. Single-equation and systems procedures are employed to estimate the parameters of the first two moments of the distribution of crop yields. The analysis supports the existence of aggregate indices for weeding and harvesting labour for oonon and sorghum, while the hypothesis of separability in hired and family labour is rejected. Tue form of labour contract for hired labour is found to have significant implications on its production risk effects. When hired labour is paid in cash, production risks increase, as is the case with cotton and sorghum. When sharecropping takes place, as in groundnuts, production risks decrease with increased labour use. Supply behaviour of the tenant farmers under production uncertainty is simulated using a farm programming model.