Despite the numerous studies on human lifting ability no Berious attempt has been made to devise an empirical prediction model based on a large number of task factors rather than individual factors. This paper first demonstrates that the effect of one variable (age) is similar in a study of 45 subjects using a non-compact object to that found in previous studies using compact objects. A model is proposed to predict lifting performance from a large number of task-related variables. A test of the model using the effects of five variables estimated from the literature was conducted. Twenty-six lifting conditions were tested and the model predictions matched the actual weights lifted with a correlation coefficient of 0-936. The implications for future extension of the model are discussed.