We conducted a series of field experiments to investigate the ability of experimentally measured risk preferences to predict the contractual choices of workers in the real labour market. In a first set of experiments we measured workers’ risk preferences using the lottery approach of Holt and Laury (2002). We did this twice: once for low-stakes lotteries (LSL) and once for high-stakes (HSL). Each worker subsequently made 12 decisions, choosing between his/her regular piece-rate contract and a series of fixed wage contracts, each offereing a different fixed wage. One of the twelve decisions was then chosen at random and the worker was paid according to his/her choice for that decision over a period of two working days. The risk preferences measured from the HSL effectively predict the contract choices; those from the LSL are irrelevant. We also find that high-ability workers prefer piece-rate contracts.