Aims There is a long-acknowledged tradeoff between precision and cost of exposure assessment strategies (Winkel and Mathiassen, 1994), but a dearth of literature quantifying this tradeoff. This study compares different sampling strategies for trunk posture observation with respect to monetary costs and statistical efficiency. Methods Experts observed 126 workers in heavy industry during full work shifts, with repeated measures on 76% of workers. The observed percentage of time spent with trunk flexed more than 60 degrees was recorded and summarized for each work day using the Back-EST sampling method (Village et al 2009). A model of costs associated with the data sampling strategy was developed using previously published data (Trask et al 2007), accounting for the costs of recruiting companies and workers, and of observing full work shifts. Statistical performance was described as standard error of the mean (SEM). Post hoc, a selection of sampling strategies were simulated to demonstrate the empirical trade-off between cost and efficiency for different sampling strategies. Results: Figure 1 shows the family of curves resulting when 1 to 10 workers are measured 1 to 4 times within a single company. The SEM declines steeply for the first few additional subjects, while further subjects increase costs considerably with no substantial improvements in precision. Adding repeated measures generally increased costs with smaller gains in precision. Discussion/conclusions: This study informs design of posture observation sampling campaigns; the ‘mesh’ of sampling strategies in the figure demonstrates the maximum precision level that can be achieved for a given cost, as well as the minimum cost for a given level of precision. This allows researchers to make informed decisions on the use of limited resources when designing ergonomic studies