The purpose of this study was twofold: (1) to assess the reliability of both the psychophysical method in determining maximum acceptable loads (MAL) and the use of selected gait parameters in describing locomotor patterns; and (2) to describe and compare the locomotor patterns associated with a number of common load carriage strategies involving the hands and arms. A test-retest experimental design was utilized. Ten males performed five tasks comprising four load carriage conditions and one normal walking condition over a 10m walking distance. The MAL for each carriage task was determined using procedures modified from those described by Snook et al. (1970). Preferred walking speed was calculated from the time taken to cover a standardized portion of the walking distance. The other gait parameters-stride length, cadence and total and double support periods-were obtained using a footswitch technology as described by Wall et al. (1981). The results demonstrated that the procedures used in this study were reliable, in that no significant test-retest differences were found for any of the dependent variables (MAL and gait parameters). No significant differences in MAL were found between the carriage strategies involving two hands, but significantly less mass was chosen in the one-hand carriage condition. Only the two-handed arms-straight carriage condition resulted in significantly different preferred walking speeds. All load carriage conditions except the one-hand carriage condition resulted in significant changes in stride length and/or cadence when compared with normal walking. Only one condition, involving the heaviest MAL, resulting in significantly different total and double support periods when compared with normal walking. There appear to be discrepancies in the literature regarding the effects of load carriage on the various gait parameters. Specific task characteristics, such as the method of carriage, load mass, speed and walking medium, were discussed as potential contributors to the seemingly contradictory results. Subject characteristics, such as gender, age and occupation, may also be confounding factors.